Skip to content
CAI
Software that uses CAICheck a score

0xPlaygrounds/rig

72.3

Strong · 14 November 2026

553k

lines of production code

Rust

primary language

5

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

AWS Bedrock provider integration with Converse API support

The \rig-bedrock\ crate now provides a complete integration with AWS Bedrock, enabling text completion, streaming, embeddings, and image generation. Users can create models via \BedrockRuntime\ using the Converse API, which supports structured features like prompt caching, guardrails, and reasoning content (including redacted reasoning blocks). The provider exposes a wide range of model identifiers (e.g., Amazon Nova, Anthropic Claude, Meta Llama, Mistral) and handles streaming events, usage tracking, and provider-specific request IDs. Also in this area: Reasoning content blobs are now captured for replay; Model catalog now drives encoding and pricing via ModelFacts; Standardize document text handling and refine error classification; Bedrock and Candle providers now support typed generation options and provider extensions; AWS Bedrock provider now supports structured output and detailed reasoning content.

crates/rig-bedrock/src · high confidence

Add Cohere image embedding example

Added a new example demonstrating how to embed images using the Cohere Embed v3 model. The example reads an image file (PNG, JPEG, WebP, or GIF) from the command line, uses the provider's image embedding capability, and prints the resulting embedding dimensions.

_examples/cohere\_image\embeddings · high confidence

Add MongoDB vector search example

A new example demonstrating how to use the rig-mongodb crate for vector search has been added. It shows how to initialize a MongoDB client, generate embeddings using OpenAI's text-embedding-ada-002 model, insert documents into a MongoDB collection, and perform a vector similarity search using a pre-existing vector index.

crates/rig-mongodb/examples · high confidence

Add Qdrant vector search example with filtering support

A new example demonstrating how to use the rig-qdrant integration for vector search has been added. It shows how to connect to a local Qdrant instance, create a collection, embed documents using OpenAI's text-embedding-ada-002 model, and perform both standard and filtered vector searches using QdrantFilter.

crates/rig-qdrant/examples · high confidence

Add Qdrant vector store integration with metadata filtering

Introduces the \rig-qdrant\ crate, providing a \QdrantVectorStore\ implementation for dense vector search against Qdrant collections. This change adds support for filtering search results using \QdrantFilter\, which allows users to apply conditions such as equality, range comparisons (exclusive/inclusive), existence checks, and logical combinations (AND/OR) to narrow down retrieved documents.

crates/rig-qdrant/src · high confidence

Add ScyllaDB vector store implementation

Introduces a new ScyllaDB-backed vector store for Rig, enabling users to store and query document embeddings using ScyllaDB. The implementation supports client-side cosine similarity scoring and includes a filter system that translates JSON-based search filters into CQL queries with proper parameter binding. Also in this area: New vector search examples for ScyllaDB and SurrealDB; Added tests for audio generation, completion, and effect bus components.

crates/rig-scylladb · high confidence

Add SurrealDB vector store integration

Users can now store and search embeddings using a SurrealDB-backed vector store. The new \SurrealVectorStore\ supports multiple distance metrics (Cosine, KNN, Euclidean, Hamming, Jaccard) and allows filtering search results using a \SurrealSearchFilter\ that translates dynamic filters into SurrealQL predicates. The implementation includes in-memory and WebSocket engine support and is covered by tests for filter conversion and type-erased query compatibility.

crates/rig-surrealdb · high confidence

Add autonomous agent example demonstrating self-correcting extraction loop

A new example has been added to the \examples/agent\_autonomous\ directory that demonstrates an autonomous agent pattern. The code implements a loop where an OpenAI-based extractor repeatedly processes its own output (incrementing a counter) until a target value is reached, showcasing how to chain extraction results back into the input stream for iterative tasks.

_examples/agent\autonomous · high confidence

Add extractor example demonstrating typed extraction and usage metadata

A new example in examples/extractor demonstrates how to use the ExtractorBuilder to perform structured data extraction from text. The example shows extracting a Person struct from input strings and highlights the ability to access usage metadata, such as total tokens, from the extraction response.

examples/extractor · high confidence

Add formatted display utility for Neo4j search results

A new \lib.rs\ file in the \crates/rig-neo4j/examples/display\ directory introduces a \SearchResults\ struct that implements the \Display\ trait. This allows users to print Neo4j vector search results in a formatted, column-aligned table (Title, ID, Description) that automatically wraps text to fit the terminal width.

crates/rig-neo4j/examples/display · high confidence

Add local inference example using rig-candle

A new example application (\candle\_local\) demonstrates running local inference via the \rig-candle\ integration. It downloads a pinned SmolLM2-360M-Instruct model (Q4\_K\_M GGUF) and executes completions through the root \rig\ facade, printing streaming text and detailed generation metrics (prefill time, throughput, etc.) to the console.

_examples/candle\local · high confidence

Add multi-agent example with translator tool

A new example demonstrating a multi-agent architecture has been added to the examples/multi\_agent directory. This sample application shows how to wrap an agent as a tool, allowing a primary agent to delegate translation tasks to a specialized translator agent before generating a final response.

_examples/multi\agent · high confidence

Add retrieval-augmented generation example

A new example demonstrating retrieval-augmented prompting has been added to the examples/chain directory. This example shows how to use an in-memory vector store to look up context (specifically, definitions of non-standard words) and fold that context into a prompt before sending it to an OpenAI agent, replacing the previous parallel pipeline approach with a clearer sequential lookup.

examples/chain · high confidence

Add shared example utilities for Bedrock integration

A new \common\ module has been added to the \rig-bedrock\ examples, providing reusable components for demonstration purposes. This includes a \MathError\ type for handling calculation failures and an \Adder\ tool implementation that demonstrates how to define a tool with specific arguments, parameters, and execution logic within the \rig-agent\ framework.

crates/rig-bedrock/examples/common · high confidence

Add vector search example using in-memory store and OpenAI embeddings

A new example demonstrates how to embed documents and query an in-memory vector index using the OpenAI provider. It shows how to define a data structure with the \\#\[embed\]\ attribute, build embeddings, and perform vector searches to retrieve both full document results and ID-only results.

_examples/vector\search · high confidence

Added Milvus vector search example

The repository now includes a new example demonstrating how to perform vector search using the Milvus vector store. This example shows users how to configure the Milvus client with authentication, generate embeddings via OpenAI, insert documents into a collection, and execute similarity searches using the rig-core vector store interface.

crates/rig-milvus/examples · high confidence

Added example fixtures for LanceDB integration

A new \lib.rs\ file has been added to the \crates/rig-lancedb/examples/fixtures\ directory, providing sample data structures and utilities for testing the LanceDB integration. This includes a \Word\ struct with an embedded definition field and helper functions to convert these words and their embeddings into Arrow \RecordBatch\ objects, facilitating easier setup for example applications.

crates/rig-lancedb/examples/fixtures · high confidence

Added vector search examples using Fastembed

New example files demonstrate how to perform vector search using the Fastembed library. The \vector\_search\_fastembed.rs\ example shows how to load a pre-built model (AllMiniLML6V2Q) and use the \EmbeddingsBuilder\ to create embeddings for documents, store them in an in-memory vector store, and query for similar results. The \vector\_search\_fastembed\_local.rs\ example illustrates how to load a custom local ONNX model by specifying model files and tokenizer configurations, then use it for embedding and search.

crates/rig-fastembed/examples · high confidence

Conversation memory with compaction, blob storage, and database-backed sessions

The rig-memory crate now provides a complete session management system for long-running agents. It introduces a \Compactor\ trait and \CompactingMemory\ adapter to automatically summarize and replace older conversation history, keeping the context window within model limits while preserving recent turns. Image data is now stored as deduplicated blobs (SHA-256 named files) rather than inline base64, reducing message size. The crate also adds database-backed session stores for SQLite and PostgreSQL, allowing conversations to be persisted beyond local memory. Additionally, it includes a \ClearToolOutputs\ policy to free context by truncating large tool results, and a conformance test suite to ensure all store implementations correctly handle concurrent writes and log integrity. Also in this area: 5 commits (1 fix) modifying crates/rig-memory; 1 commit (0 fixes) modifying crates/rig-memory, plugins/rig-traffic; 1 commit (1 fix) fixing crates/rig-memory.

crates/rig-memory/src · medium confidence · unverified

Example demonstrating Gemini API recovery via custom hooks

The \examples/gemini\_default\_api\_recovery\ example now shows how to handle legacy tool name emissions from the Gemini provider. It implements a custom \AgentHook\ (\DefaultApiRepairHook\) that intercepts invalid tool calls, specifically mapping the legacy \default\_api\ tool name to the current \JavaScript\ tool, allowing the agent to recover gracefully from provider-specific naming quirks.

_examples/gemini\_default\_api\recovery · high confidence

Introduce AI coding agent guidelines and pre-commit hooks

Added AGENTS.md to provide operational instructions for AI coding agents, including rules on code style, error handling, and provider implementation. Configured .pre-commit-config.yaml to enforce trailing whitespace, YAML/JSON checks, Rust formatting, and commit message conventions via commitizen.

(repo-wide) · high confidence

Introduce AWS S3Vectors vector store integration

Users can now store and search vectors using AWS S3Vectors. This new \rig-s3vectors\ crate provides an \S3VectorsVectorStore\ implementation that integrates with the AWS SDK, supporting document insertion and search with dynamic filtering (including equality, range, existence, and boolean logic) via \S3SearchFilter\. The implementation includes tests verifying that dynamic filters correctly compile to native AWS document structures. Also in this area: New vector search examples for Postgres and S3 Vectors; Added tests for audio generation, completion, and effect bus components.

crates/rig-s3vectors · high confidence

Introduce Milvus vector store integration with typed filter expressions

Added a new Milvus vector store implementation in \crates/rig-milvus\ that enables inserting and searching documents via the Milvus v2 HTTP API. The change introduces a \Filter\ type and \MilvusValue\ enum to construct typed, escaped boolean filter expressions (supporting equality, comparison, logical AND/OR, \IN\, \LIKE\, and array containment operations) for precise query filtering. The store handles authentication via Bearer tokens, manages embedding generation using a provided model, and serializes search requests to the Milvus endpoint, returning results with optional ID-only or full document payloads.

crates/rig-milvus/src · high confidence

Introduce Neo4j vector store integration

Added the \rig-neo4j\ crate, providing a \Neo4jVectorIndex\ that enables vector similarity search over Neo4j graph databases. Users can now query existing Neo4j vector indexes using configurable similarity functions (Cosine or Euclidean) and apply rich search filters (equality, range, containment, and regex) via the \Neo4jSearchFilter\ API. The implementation supports inserting documents into specific node labels and includes tests for configuration defaults and query generation.

crates/rig-helixdb/src, crates/rig-mongodb/src, crates/rig-neo4j/src · high confidence

Introduce \`rig\` facade crate for unified API access

A new \src/lib.rs\ file establishes the \rig\ crate as a public facade that re-exports core types from \rig\_core\ and agent-specific runtime components from \rig\_agent\ (such as \Agent\, \AgentRun\, and \ToolSet\) behind feature flags. This change provides a single entry point for users to access portable provider contracts, transport-agnostic HTTP clients, and classic agent orchestration APIs, while also exposing optional integrations like the \reqwest\ transport, \tungstenite\ websockets, and MCP tool support via specific features.

src · high confidence

Introduce audio generation (TTS) support and provider client abstractions

Users can now generate audio from text using the new \AudioGenerationRequestBuilder\ and \AudioGenerationResponse\ types, which expose provider metadata, usage counters, and response identity. The \rig-core\ client layer has been restructured to use a shared \http\_client!\ macro, providing a uniform pattern for provider clients (Anthropic, Cohere, Copilot, Gemini, Ollama, OpenAI, Voyage AI) that manage configuration, transport, and model instantiation. Additionally, Gemini-specific explicit context caching operations are now exposed through the client layer. Also in this area: 19 commits (6 fixes) modifying crates/rig-core; Aborted replies now replay as partial text, and ChatGPT reasoning effort requests auto summaries; Recording proxy for provider traffic and effect record simplification; Added test fixtures for DeepSeek strict tools and provider responses routes; Model catalog now drives encoding and pricing via ModelFacts; and 3 more.

(repo-wide) · medium confidence · unverified

Introduce local embedding support via fastembed integration

The new \rig-fastembed\ crate provides a local embedding transport backed by the \fastembed\ library, allowing users to run text embeddings directly in the calling process without external services. It exposes a \Fastembed\ struct for loading models (including automatic downloads via Hugging Face Hub when the \hf-hub\ feature is enabled) and a \text\_embeddings\ wire function to configure embedding capabilities. This enables in-process vectorization for supported models like \AllMiniLML6V2Q\, with errors handled via a dedicated \FastembedError\ type.

crates/rig-fastembed/src · high confidence

Introduction of rig-cassette crate for recording and replay

The new rig-cassette crate provides recording and replay capabilities for Rig's effects and provider HTTP exchanges. It exposes an \effect\_log\ module by default (supporting WASM) and offers optional feature-gated modules for agent integrations (\agent\), ECS checkpoint boundaries (\ecs\), and native HTTP engine cassettes with ordered JSON maps and round-trip float parsing (\http\), which can be further extended with AWS event-stream support via the \bedrock\ feature. Also in this area: Updated ChatGPT cassette fixtures to reflect API changes; Recording proxy for provider traffic and effect record simplification; Standardize document text handling and refine error classification.

crates/rig-cassette/src · high confidence

LanceDB integration now converts query results to JSON and filters embedding columns

The \rig-lancedb\ crate now includes utilities to execute LanceDB queries and automatically convert the resulting Arrow record batches into JSON rows, supporting a wide range of data types including primitives, strings, binary data, nested lists, structs, maps, and dictionaries. Additionally, a schema filtering utility is provided to automatically exclude \FixedSizeList\ columns containing \Float64\ items (used for embeddings) from query projections, ensuring that only relevant data columns are returned to the user.

crates/rig-lancedb/src/utils · high confidence

Local CPU inference for LLMs and YOLOv8 pose estimation

The rig-candle crate now enables running models directly on the CPU without external services. It supports text completion for Llama, SmolLM2, and Qwen3 architectures using both safetensors and GGUF checkpoint formats, with configurable generation parameters like temperature and top-k. Additionally, it introduces a new YOLOv8 pose estimation capability that accepts RGB image frames to detect and track human poses locally. Also in this area: 1 commit (1 fix) fixing crates/rig-candle; 1 commit (0 fixes) modifying (repo-wide); Model facts now drive pricing and option rules across all provider wires; Agent engine ensures one-to-one completion dispatch/outcome and per-turn tool choice policy; Bedrock and Candle providers now support typed generation options and provider extensions.

crates/rig-candle/src · low confidence · unverified

New \#\[rig\_tool\] macro with strict argument grammar and context parameter handling

The \crates/rig-derive/src/tool\ module introduces the implementation for the \\#\[rig\_tool\]\ attribute macro. It enforces a strict grammar for macro arguments (name, description, params, required), rejecting duplicates and invalid formats. It also classifies function parameters to identify and handle a mutable \ToolContext\ parameter (marked via \\#\[rig(context)\]\ or qualified type), ensuring it is excluded from the generated JSON schema while being passed to the underlying function.

crates/rig-derive/src/tool · high confidence

New AWS Bedrock example applications

Added four new example applications demonstrating AWS Bedrock integration: agent\_with\_bedrock.rs shows basic agents, tool usage, context injection, and file loading; embedding\_with\_bedrock.rs demonstrates text embedding generation; image\_generator.rs illustrates image generation capabilities; and streaming\_with\_bedrock\_and\_tools.rs provides an example of streaming responses with tool use.

crates/rig-bedrock/examples · high confidence

New Bevy-based agent example and updated Candle local inference API

The \examples/agent\_no\_tokio\ directory now provides a new example demonstrating how to run a rig agent on Bevy's \AsyncComputeTaskPool\ without requiring tokio in the crate's manifest, utilizing \AgentRunner::run\_channel\ to split execution into a spawned future and a non-blocking event feed. Additionally, the \examples/candle\_local\ example has been updated to reflect API changes: it now uses \CompletionRequest\ for building requests and \CandleModel::stream\ for streaming, replacing the previous \raw\_stream\ approach, while still accessing Candle-specific metrics via the response's raw document.

_examples/agent\_no\tokio · high confidence

New CLI chatbot integration for interactive agent sessions

Added a new \cli\_chatbot\ module that provides a terminal-based chat loop for interacting with agents. Users can now use \ChatBotBuilder\ to configure and run multi-turn conversations with streaming output, optional usage reporting, and history management, either with the default \Agent\ implementation or custom types implementing the \Chat\ trait.

crates/rig-agent/src/integrations · high confidence

New Gemini Deep Research example demonstrating agent-based interactions and streaming

Added a new example in \examples/gemini\_deep\_research\ that showcases the Gemini provider's Deep Research capabilities. The example demonstrates how to configure a specific research agent via environment variables, construct requests using \additional\_params\ for agent-specific settings like thinking summaries, and handle the resulting interaction lifecycle. It supports both synchronous polling for terminal states and asynchronous streaming, correctly parsing and displaying reasoning steps and final text outputs from the provider's interaction model.

_examples/gemini\_deep\research · high confidence

New Gemini gRPC agent example

Added a new example demonstrating how to use the Gemini gRPC provider with the rig-agent facade. The example shows creating a GeminiGrpc transport from environment variables, building an agent with specific model and temperature settings, and executing a prompt.

crates/rig-gemini-grpc/examples · high confidence

New Gemini gRPC transport for completions and embeddings

Users can now connect to Google Gemini models via a gRPC transport instead of the previous REST-based implementation. This new \rig-gemini-grpc\ crate provides a \GeminiGrpc\ client that handles API key authentication and TLS, exposing both completion (\GenerateContent\) and embedding (\EmbedContent\) capabilities. The transport supports both unary and streaming modes, correctly decodes streaming responses into normalized completion events (including reasoning/thought blocks and tool calls), and maps gRPC status codes to provider errors for proper retry and error handling.

crates/rig-gemini-grpc/src · high confidence

New Gemini video understanding example with custom generation config

Added a new example demonstrating how to use the Gemini API for video understanding. The example shows how to construct a user message containing both text instructions and a video URL, and how to pass provider-specific parameters like \GenerationConfig\ (e.g., \top\_k\, \top\_p\) via \AdditionalParameters\ when building the agent.

_examples/gemini\_video\understanding · high confidence

New HTTP middleware example for transport-level request/response inspection

An example demonstrating how to attach an HTTP middleware to the provider's client to intercept and log wire-level details. The \WireLogger\ middleware injects a specific \anthropic-beta\ header before requests, logs the HTTP method, URI, and payload size, and reads response headers (including rate-limit and request-id metadata) before the response body is consumed.

_examples/http\middleware · high confidence

New LanceDB vector search examples for local, S3, and agent integration

Added four new Rust examples in \crates/rig-lancedb/examples\ demonstrating how to use the \rig-lancedb\ crate for vector search. \vector\_search\_local\_ann.rs\ and \vector\_search\_local\_enn.rs\ show how to perform approximate and exact nearest-neighbor searches using a local LanceDB instance with OpenAI embeddings. \vector\_search\_s3\_ann.rs\ extends this to demonstrate storing and searching vector data in AWS S3 with Cosine distance. \vector\_search\_local\_ann\_agent.rs\ illustrates building a RAG agent using \rig\_agent::AgentBuilder\ with dynamic context sourced from a LanceDB vector index.

crates/rig-lancedb/examples · high confidence

New LanceDB vector store integration

Users can now store and search vectors using LanceDB via the new \LanceDbVectorIndex\. This implementation supports both flat and approximate nearest neighbor search strategies, allows configuration of distance metrics and search parameters (such as nprobes and refine factor), and enables filtering results using SQL-like predicates through the \LanceDBFilter\ type.

crates/rig-lancedb/src · high confidence

New Neo4j vector search examples for embedding and querying

Added three new examples in the \rig-neo4j\ crate demonstrating how to use Neo4j as a vector store: \vector\_search\_simple.rs\ shows a basic end-to-end flow of generating embeddings, creating a vector index, and querying it; \vector\_search\_movies\_add\_embeddings.rs\ demonstrates how to batch-generate embeddings for existing movie data and attach them to nodes; and \vector\_search\_movies\_consume.rs\ shows how to query an existing Neo4j database with pre-computed embeddings using the \VectorStoreIndex\ API.

crates/rig-neo4j/examples · high confidence

New PDF RAG example using Ollama and dynamic context

A new example application has been added at examples/pdf\_agent that demonstrates a Retrieval-Augmented Generation (RAG) workflow. The application loads and chunks PDF documents using a built-in loader, generates embeddings with the local Ollama provider (bge-m3 model), stores them in an in-memory vector store, and uses a dynamic context helper to inject relevant document chunks into a deepseek-r1 agent for answering questions via a CLI chatbot interface.

_examples/pdf\agent · high confidence

New RAG examples for Gemini, OpenAI, and Ollama

Added three new example applications demonstrating Retrieval-Augmented Generation (RAG) patterns using the \rig\ library. The \gemini\_extractor\_with\_rag\ example shows how to use the \ExtractorBuilder\ with Gemini to parse structured questionnaire responses from unstructured text. The \rag\ and \rag\_ollama\ examples demonstrate building agents with \AgentBuilder\ that use dynamic context (vector search) to answer questions based on a local in-memory vector store, supporting both OpenAI and Ollama providers.

_examples/gemini\_extractor\_with\rag · high confidence

New RAG examples for Gemini, OpenAI, and Ollama

Added three new example applications demonstrating Retrieval-Augmented Generation (RAG) patterns using the \rig\ library. The \gemini\_extractor\_with\_rag\ example shows how to use the \ExtractorBuilder\ with Gemini to parse structured questionnaire responses from unstructured text. The \rag\ and \rag\_ollama\ examples demonstrate building agents with \AgentBuilder\ that use dynamic context (vector search) to answer questions based on a local in-memory vector store, supporting both OpenAI and Ollama providers.

examples/rag · high confidence

New RMCP example demonstrating MCP server capabilities

Added a new example in \examples/rmcp/src/main.rs\ that shows how to build an MCP-compatible agent using the \rmcp\ library. The example implements a \Counter\ server that exposes tools (such as a sum calculator) and resources (like a current working directory and a memo), demonstrating both static tool fetching and auto-updating via \McpClientHandler\ notifications. It also illustrates handling HTTP transport via \StreamableHttpService\ and integrating with the \rig\ framework's provider and tool abstractions.

examples/rmcp · high confidence

New agent and extraction examples added to the examples directory

The examples/agent location now includes three new Rust source files demonstrating core library capabilities: a minimal agent setup using OpenAI's GPT-4o, an example showing how to configure OpenAI's Chat Completions API with OpenTelemetry tracing, and a typed sentiment classification example using the ExtractorBuilder. These files provide concrete usage patterns for building agents, configuring API routes, and performing structured data extraction.

examples/agent · high confidence

New agent examples demonstrate raw response access, truncation retry, and runtime model routing

Three new examples in the \rig-agent\ crate illustrate key agent capabilities: \raw\_response\_hook.rs\ shows how to access provider-specific fields (like OpenAI's \system\_fingerprint\) via the \CompletionResponse::raw\ surface in hooks; \retry\_on\_truncation.rs\ demonstrates a provider-neutral strategy to automatically retry turns when the provider truncates output based on \FinishReason\ and \max\_tokens\; and \runtime\_model\_routing.rs\ shows how to route an agent across different models (e.g., a 'fast' research model and a 'strong' synthesis model) at runtime using a \ModelSelection\ hook. Also in this area: Bedrock and Candle providers now support typed generation options and provider extensions.

crates/rig-agent/examples · high confidence

New agent pattern examples: prompt chaining and routing

Added two new Rust examples demonstrating multi-agent workflows. The \agent\_prompt\_chaining\ example shows how to sequence agents, where the output of one agent serves as the input for the next. The \agent\_routing\ example demonstrates a classifier agent that directs a prompt to different follow-up handlers based on its categorization. Both examples use the updated API where the model type is erased at construction (\erase()\) and shared via cloned handles.

_examples/agent\_prompt\_chaining, examples/agent\routing · high confidence

New agent streaming example with conversation history

The examples/agent\_stream\_chat directory now contains a new Rust example demonstrating how to stream agent responses while maintaining conversation history. Users can see how to initialize an agent, pass prior messages via the history method, and consume the resulting stream to extract the final response.

_examples/agent\_stream\chat · high confidence

New cassette maintenance CLI tool for account failures, stored state, and cleanup

A new example tool (\cassette\_tool\) is available to help maintain cassette integrity. It provides three commands: \account-failures\ to detect replies that would be refused as undeclared account failures, \stored-state\ to identify cassettes storing provider response state without cleanup, and \cleanup\ to delete resources tracked in the created-resource ledger. These utilities allow users to audit and clean up recording artifacts directly from the command line.

crates/rig-cassette/examples · high confidence

New debate example demonstrating multi-agent interaction

Added a new example in the \examples/debate\ directory that showcases a simulated debate between two AI agents. The example instantiates one agent using OpenAI's GPT-4 and another using Cohere's Command model, allowing them to exchange arguments over multiple rounds while maintaining separate conversation histories.

examples/debate · high confidence

New derive macros for tool definitions and context values

The \rig-derive\ crate now provides \\#\[derive(ContextValue)\]\ to automatically generate slot keys for tool context values, and \\#\[rig\_tool\]\ to generate tool types, parameter structs, and metadata from functions. These macros simplify the creation of tools by handling schema generation and context resolution, allowing developers to define tools with minimal boilerplate while ensuring proper integration with the core tooling infrastructure.

crates/rig-derive/src · high confidence

New durable human-in-the-loop approval example

Added a new example demonstrating durable human-in-the-loop tool-call approval, where the agent's state is serialized to a JSON file between turns to allow decisions to be made asynchronously or across process boundaries. The example implements a fail-closed approval gate for a side-effecting 'transfer\_funds' tool, supporting approve, deny, edit, and abort actions, while using a read-only 'get\_balance' tool. Also in this area: Fix: Agent runs now survive restarts with resumable in-flight tool calls; Update examples to use TurnPolicy for tool choice management.

_examples/agent\_with\_durable\approval · high confidence

New example demonstrating agent self-correction via evaluator feedback

Added a new example in \examples/agent\_evaluator\_optimizer\ that illustrates a loop where a generator agent produces code and an evaluator agent, using an \ExtractorBuilder\ to parse structured JSON responses, provides feedback. The generator then uses this feedback to refine its output until the evaluator marks the status as 'Pass', demonstrating a practical pattern for iterative agent improvement.

_examples/agent\_evaluator\optimizer · high confidence

New example demonstrating complex agentic loops with Claude

Added a new example in \examples/complex\_agentic\_loop\_claude\ that showcases a multi-agent orchestration pattern using the Anthropic Claude model. The example implements an orchestrator agent that utilizes specialized sub-agents (research, data analysis, and recommendation) as dynamic tools, alongside a vector store for knowledge retrieval and a think tool for complex reasoning. It demonstrates how to configure an agent to perform multi-turn interactions with a maximum of 15 turns, illustrating a sophisticated agentic workflow for environmental sustainability advice.

_examples/complex\_agentic\_loop\claude · high confidence

New example demonstrating composable agent hooks with merged request patches

The \examples/request\_hook\ example now illustrates the v2 hook system, showing how multiple \AgentHook\ implementations (logging, context injection, temperature adjustment, and turn counting) can be stacked via \.add\_hook()\. It demonstrates that request patches from different hooks merge rather than short-circuit, allowing combined effects like adding context and lowering temperature on the same turn, while using a shared scratchpad for state management.

_examples/request\hook · high confidence

New example demonstrating context injection in agents

Added a new example in \examples/agent\_with\_context\ that shows how to build an agent using the Cohere provider and inject specific context documents (definitions of terms like 'flurbo' and 'glarb-glarb') directly into the agent's builder before prompting.

_examples/agent\_with\context · high confidence

New example demonstrating custom reqwest middleware with retry logic

Added a new example in \examples/reqwest\_middleware\ that shows how to bind a preconfigured HTTP client to an agent. The example uses \reqwest-middleware\ and \reqwest-retry\ to apply exponential backoff retry policies to Anthropic API calls, illustrating how users can inject custom transport behavior into the rig agent.

_examples/reqwest\middleware · high confidence

New example demonstrating how to correctly force a tool call on the first turn only

Added an example in \examples/force\_tool\_first\_turn\ that illustrates a common pitfall when using \RequestPatch\ to force tool usage: applying \tool\_choice = Required\ unconditionally causes the model to loop indefinitely until it hits the \max\_turns\ limit. The example contrasts this 'footgun' with the correct approach—gating the patch on the first turn (\ctx.turn() == 1\)—so the model is nudged to call the tool initially but can then stop and provide a final answer.

_examples/force\_tool\_first\turn · high confidence

New example demonstrating manual tool-call handling

Added a new example in \examples/manual\_tool\_calls\ that shows how to handle tool calls manually using a raw \Model\ request instead of the high-level \Agent\ abstraction. The example implements a calculator loop that sends completion requests, collects \ToolCall\s, executes them locally via a \ToolSet\, and feeds results back to the model until a final text answer is returned.

_examples/manual\_tool\calls · high confidence

New example demonstrating per-call usage inspection in agent streams

Added an example (\openai\_streaming\_per\_call\_usage\) that shows how to inspect token usage for individual completion calls within an agent's multi-turn stream, rather than relying solely on the aggregated final response usage. The example illustrates handling \MultiTurnStreamItem::CompletionCall\ events to retrieve input and output token counts for each step (such as tool requests and final answers), which is useful for understanding context size and cost distribution during complex agent runs.

_examples/openai\_streaming\_per\_call\usage · high confidence

New example demonstrating persistent multi-turn conversation memory

Added an example in \examples/agent\_with\_memory\ that shows how to maintain conversation context across multiple prompts using \rig::memory\. The example initializes a \MemoryStore\, loads a conversation history for a specific user, and iteratively sends prompts to an OpenAI agent while appending the resulting messages back to the store. It supports both standard synchronous responses and streaming responses (via the \--stream\ flag), ensuring that the agent retains context from previous turns in subsequent interactions.

_examples/agent\_with\memory · high confidence

New example demonstrating structured tool failure handling and policy hooks

Added a new example (\tool\_result\_outcomes\) that demonstrates how to classify tool execution errors (such as disk I/O or network failures) into structured facts and apply specific policies via hooks. The example shows how to use \OutcomeEvent\ to record typed metadata in a run-scoped scratchpad, allowing one hook to record failure details and a subsequent hook to decide whether to terminate the run or allow the model to retry, distinguishing between fatal and recoverable errors.

_examples/tool\_result\outcomes · high confidence

New example demonstrating token-count estimation for disrupted Gemini streams

Added a new example (\gemini\_stream\_kill\_token\_count\) that illustrates how to estimate token usage when a Gemini streaming generation is interrupted mid-response. Since the API only provides authoritative usage metadata on the final chunk, this example implements a fallback mechanism that uses the \countTokens\ endpoint on partial output if the stream is killed, encounters a transport error, closes prematurely, or stalls. It injects four distinct disruption scenarios to demonstrate a unified accounting path for handling incomplete streams.

_examples/gemini\_stream\_kill\_token\count · high confidence

New examples demonstrating agent stepping and EchoChambers integration

Added two new example applications: \agent\_run\_stepping\ shows how to manually drive the \AgentRun\ state machine for granular control (including serialization/pause/resume) and how to use the high-level \AgentRunner\ with hooks, while \agent\_with\_echochambers\ demonstrates an agent interacting with the EchoChambers API via custom tools for sending messages, retrieving history, and fetching room metrics.

_examples/agent\_run\stepping · high confidence

New examples demonstrating parallel structured extraction

Added \agent\_parallelization\ and \multi\_extract\ examples to the \examples\ directory. These demonstrate how to use \ExtractorBuilder\ with \futures::join!\ and \futures::try\_join!\ to perform concurrent, fan-out structured extraction tasks, such as scoring text on multiple dimensions or extracting names, topics, and sentiment in parallel.

_examples/multi\extract · high confidence

New examples demonstrating policy-based and interactive human-in-the-loop tool approval

Added two new example applications in the \examples/agent\_with\_approval\_policy\ and \examples/agent\_with\_human\_in\_the\_loop\ directories that showcase the \AgentHook\ system for controlling tool execution. The \agent\_with\_approval\_policy\ example implements a non-interactive, fail-closed policy where specific tools (like \search\_web\) are auto-approved while others (like \transfer\_funds\) are gated by rules (e.g., amount limits), with denials fed back to the model as tool results. The \agent\_with\_human\_in\_the\_loop\ example demonstrates an interactive approval flow where the agent pauses on tool calls (like \send\_email\ or \delete\_file\) to wait for human input via stdin, supporting actions to approve, deny, edit arguments, or abort the run entirely.

_examples/agent\_with\_approval\policy · high confidence

New examples for Gemini image generation and multi-provider transcription

Added two new example applications to demonstrate specific capabilities. The \gemini\_nanobanana\_image\_generation\ example shows how to use the Gemini provider to generate images (specifically a 512x512 banana icon) using the new \ImageGenerationRequestBuilder\. The \transcription\ example demonstrates audio transcription across multiple providers (OpenAI Whisper, Gemini, Azure, Groq, HuggingFace, and Mistral) using a unified \TranscriptionRequestBuilder\ interface.

_examples/gemini\_nanobanana\_image\generation · high confidence

New examples for agent tool usage and turn budgets

Added two new example programs in the \examples/agent\_with\_tools\ area: \agent\_with\_default\_max\_turns\ demonstrates configuring a default turn budget for tool-heavy prompts using static tools, while \agent\_with\_tools\ shows how to register dynamic, runtime-defined tools with an agent.

_examples/agent\_with\tools · high confidence

Added two new example programs to demonstrate vector search capabilities: \custom\_vector\_store\ shows how to implement the \VectorStoreIndex\ trait for a Redis backend using OpenAI embeddings, and \vector\_search\_cohere\ demonstrates semantic search with Cohere by configuring separate document and query embedding models.

_examples/custom\_vector\store · high confidence

New examples for the rig\_tool macro demonstrating async, complex types, and explicit metadata

Added five new examples in the rig\_tool directory that demonstrate the updated \#\[rig\_tool\] macro capabilities. The examples show how to define tools with async execution (async\_tool.rs), handle complex input types like enums and vectors (complex\_types.rs), use explicit attribute overrides for descriptions and required parameters (with\_description.rs), and build agents with multiple tools (simple.rs, full.rs). These examples illustrate the new schema generation via schemars and the simplified tool execution API.

crates/rig-derive/examples · high confidence

New examples for tool-call streaming, local inference, and WASM chat

The examples directory now includes three new demonstrations: \agent\_tool\_call\_streaming\ shows how tool-call arguments stream in real-time and how a hook can interrupt a run before a tool executes; \candle\_local\ and \candle\_pose\ provide local inference examples using rig-candle for instruction-following and YOLOv8 pose estimation respectively; and \candle\_wasm\_chat\ runs a local SmolLM2 model entirely in the browser via WebAssembly, with a Web Worker handling inference and a bounded conversation history. The examples README has been updated to document these new entries and the new package-per-example layout.

examples · high confidence

New local WASM chat example with embedded SmolLM2 model

This change introduces a new browser-based chat example in \examples/candle\_wasm\_chat\ that runs the SmolLM2-360M-Instruct model entirely within the browser via WebAssembly. The example embeds the model artifacts (config, tokenizer, and GGUF weights) directly into the WASM module, ensuring that chat messages never leave the page. It provides a complete local inference pipeline: a Rust backend (\src/lib.rs\) using the Candle library, a Web Worker to prevent UI freezing, and a JavaScript frontend (\www/app.js\) with a chat interface. The build process (\build.sh\, \download\_model.sh\) handles downloading and verifying the pinned model artifacts, while the application manages conversation history with bounded windows to fit within memory constraints.

_examples/candle\_wasm\chat · high confidence

New reasoning loop example demonstrating chain-of-thought extraction and tool use

Added a new example in \examples/reasoning\_loop\ that demonstrates a multi-step reasoning pattern. The example implements a \ReasoningAgent\ that first uses an extractor to derive reasoning steps from a user prompt, then passes those steps to an executor agent equipped with arithmetic tools (Add, Subtract, Multiply, Divide). This showcases how to combine structured extraction with tool-using agents in a single workflow.

_examples/reasoning\loop · high confidence

New release and dependency-verification tooling in scripts/

The repository now includes a suite of scripts to support the release process and ensure dependency integrity. \scripts/release-notes.sh\ gathers commit logs, merged PR details, and public API diffs into a directory, which is then processed by \scripts/release-notes-prompt.md\ to generate the \CHANGELOG.md\ and \MIGRATING.md\ files. Additionally, \scripts/check-dependency-floors.py\ verifies that the workspace builds against the lowest declared versions of all direct dependencies, with \scripts/test\_dependency\_floors.py\ providing regression tests for this verification logic.

scripts · high confidence

New rig-inspect plugin for read-only Bevy world inspection

The rig-harness coding agent now includes an optional \inspect\ tool that allows it to query its own running Bevy world via the Bevy Remote Protocol. This plugin exposes read-only methods—such as listing components, resources, schedules, and registry schemas—enabling the agent to understand its environment, plugins, and saved state without modifying anything. Write operations are strictly refused, and large responses are truncated and saved to a spill directory for the agent to read separately. The plugin is disabled by default to avoid build overhead and requires manual enabling in the plugins configuration.

plugins/rig-inspect · high confidence

New tool registry and catalog system in rig-agent

The \rig-agent\ crate now exposes a structured tool management system with \ToolServer\ for mutable, bus-published registries and \ToolCatalog\ for request-scoped, pinned tool advertisements. Users can register tools via \ToolServer\, retrieve consistent snapshots via \ToolServerHandle::snapshot\ or \static\_tool\_defs\, and execute tools through the catalog, which ensures that snapshots retain access to implementations even after the live registry is modified.

crates/rig-agent/src/tool · high confidence

New transcription example demonstrating multi-provider audio processing

The examples/transcription directory now includes a runnable main.rs that demonstrates how to transcribe audio files using the rig library. It showcases usage across multiple providers including OpenAI (Whisper-1), OpenRouter (with diarization and word timings), Gemini, Azure, Groq, HuggingFace, and Mistral. The example allows users to specify a single provider via command-line arguments or run all supported providers sequentially, illustrating the unified TranscriptionRequestBuilder interface.

examples/transcription · high confidence

New transport-agnostic HTTP client layer with middleware and streaming support

The \rig-http\ crate now provides a foundational, transport-independent HTTP client layer that replaces previous ad-hoc implementations. This introduces a type-erased \DynHttpClient\ that supports attaching middleware hooks for request/response interception, incremental framing for SSE and NDJSON streams, and a generic multipart form builder. Users gain a unified interface for HTTP interactions across different backends (like \reqwest\ or \tungstenite\) with built-in support for lazy response bodies, error preservation (including status codes and headers on failures), and runtime handling for non-Tokio environments.

crates/rig-http/src · high confidence

New typesafeai example applications for Jev evaluations

Added two new example applications, \typesafeai\_chat\ and \typesafeai\_triage\, that demonstrate the new \typesafeai\ provider integration. The chat example provides an interactive CLI for evaluating support conversations using the Jev evaluation engine, while the triage example shows how to evaluate a synthetic support ticket for routing, urgency, and duplicate charge detection.

_examples/typesafeai\chat · high confidence

New vector search examples for HelixDB and Cloudflare Vectorize

Added example applications demonstrating vector search capabilities for two new vector store integrations: HelixDB and Cloudflare Vectorize. The HelixDB example includes configuration files (schema, queries, and project settings) and a Rust application that uses OpenAI embeddings to insert and search word definitions. The Cloudflare Vectorize example provides a Rust application showing how to integrate with the Vectorize service using environment-based credentials for account ID and API token, also leveraging OpenAI embeddings for document insertion and similarity search.

crates/rig-helixdb/examples · high confidence

New vector search examples for Postgres and S3 Vectors

Added example code demonstrating vector search capabilities using the Postgres and S3 Vectors integrations. The Postgres example includes a migration script to set up a vector-enabled table and a Rust application that generates embeddings via OpenAI, inserts documents, and performs similarity searches. The S3 Vectors example provides a complete workflow for creating an S3 vector bucket and index, embedding documents with OpenAI, and querying results.

crates/rig-postgres/examples · high confidence

New word\_count tool plugin with file-size gating and persistent totals

Introduces the \WordCountPlugin\, which adds a \word\_count\ tool that counts words in workspace files. The plugin enforces a 1 MiB file size limit, refusing calls on larger files before execution, and maintains a per-agent total of words counted across sessions, accessible via the \/wc-reset\ command to zero the counter.

corpus/rf-05 · high confidence

New xtask commands for Bevy dependency validation and cassette corpus management

The xtask tooling now includes a new \bevy\ check that enforces workspace-wide Bevy dependency pinning (specifically version 0.20.0) and source restrictions, alongside a comprehensive \cassette\ command suite for managing provider test recordings. This suite introduces commands to record fixtures with attempt ledgers and caps, generate and verify acceptance indices for request facts, maintain a reply bank of real recorded responses, and regenerate effect goldens, providing developers with structured tooling to validate dependencies and manage the provider cassette corpus. Also in this area: 3 commits (1 fix) modifying xtask.

xtask · medium confidence · unverified

Rig-agent crate restructured with new extraction and completion modules

The rig-agent crate has been reorganized to provide a cleaner public API and new capabilities. A new \extractor\ module introduces an \Extractor\ and \ExtractorBuilder\ for typed structured data extraction from text, supporting configurable retries and dynamic context. The \completion\ module now centralizes runtime errors and portable completion contracts, including a \StructuredOutputError\ enum and helper macros for forwarding provider response details. The crate's public surface is updated via \lib.rs\ and \prelude.rs\ to expose these new types alongside existing agent and tooling components, while \streaming.rs\ and \sync.rs\ provide portable streaming types and synchronization primitives respectively.

crates/rig-agent/src · high confidence

Removals

Removal of legacy core modules (Agent, Model, RAG, Extractor, CLI)

The \agent\, \model\, \rag\, \extractor\, \cli\_chatbot\, \completion\, \embeddings\, and \json\_utils\ modules have been removed from \rig-core\. This eliminates the previous \Agent\ and \RagAgent\ abstractions, the synchronous \Model\ wrapper, the \Extractor\ for structured data extraction, and the built-in CLI chatbot utility, signaling a shift away from these specific legacy patterns in favor of newer architectural approaches. Also in this area: Removal of in-memory and MongoDB vector store implementations; Removal of legacy example files; Removal of legacy Cohere and OpenAI provider implementations.

rig-core · high confidence

Behavioural changes

1 commit (0 fixes) modifying corpus/rf-06

A change to existing behaviour in corpus/rf-06 — 1 commit, 3 files: “refactor(corpus): gem reads --thinking with rig-core's reasoning reader (\#3310)” (refs \#3310). Also in this area: New reference implementations for shared threads, app chat, and Gemini assistant.

corpus/rf-06 · medium confidence · unverified

10 commits (1 fix) modifying crates/rig-core/src/loaders

A change to existing behaviour in crates/rig-core/src/loaders — 10 commits (1 fix), 4 files: “chore: move every inline test module to a sibling file (\#2433)”; “docs: set doctest to true, and update doc comments (\#1716)”; “docs: finish the comment cleanup across the workspace (\#2575)”; and 7 more (refs \#2433, \#1716, \#2575, \#2409, \#2305, …).

crates/rig-core/src/loaders · medium confidence · unverified

10 commits (2 fixes) modifying crates/rig-core/src/completion/options

A change to existing behaviour in crates/rig-core/src/completion/options — 10 commits (2 fixs), 3 files: “feat(openai): Responses for MiniMax, Moonshot, Groq and Doubleword, cache breakpoints, caller-built input (\#2888)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; “refactor: provider knowledge out of generic code, typed lookups (\#2897)”; and 7 more (refs \#2888, \#2750, \#2897, \#2817, \#2838, …).

crates/rig-core/src/completion/options · medium confidence · unverified

11 commits (3 fixes) modifying crates/rig-cassette/fixtures/bank

A change to existing behaviour in crates/rig-cassette/fixtures/bank — 11 commits (3 fixs), 21 files: “feat(openai): Responses for MiniMax, Moonshot, Groq and Doubleword, cache breakpoints, caller-built input (\#2888)”; “feat(cohere)!: native documents, citations and tool\_plan (\#2729)”; “feat(gemini): Gemini 3 image, text-to-speech, current deep research (\#2809)”; and 8 more (refs \#2888, \#2729, \#2809, \#2721, \#2889, …).

crates/rig-cassette/fixtures/bank · medium confidence · unverified

12 commits (0 fixes) modifying crates/rig-core/src/test\_utils/completion

A change to existing behaviour in crates/rig-core/src/test\_utils/completion — 12 commits, 1 file: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “chore: move every inline test module to a sibling file (\#2433)”; “refactor!: archimpro — one typed decoder and one fold per wire (\#2617)”; and 9 more (refs \#2640, \#2433, \#2617, \#2535, \#2611, …).

_crates/rig-core/src/test\utils/completion · medium confidence · unverified

12 commits (1 fix) modifying crates/rig-tungstenite

A change to existing behaviour in crates/rig-tungstenite — 12 commits (1 fix), 7 files: “chore: move every inline test module to a sibling file (\#2433)”; “docs: simplify production comments and preserve contract rationale (\#2568)”; “chore: forbid inline test modules via cargo xtask check-test-layout (\#2434)”; and 9 more (refs \#2433, \#2568, \#2434, \#2567, \#2643, …).

crates/rig-tungstenite · medium confidence · unverified

12 commits (3 fixes) modifying crates/rig-core/src/completion/history

A change to existing behaviour in crates/rig-core/src/completion/history — 12 commits (3 fixs), 1 file: “fix(rig-agent)!: runs survive a restart at any point: one answer owner, resumable in-flight steps, executable-only CallTools (\#2762)”; “feat(harness): add claude-cloak and harden Anthropic handling (\#2875)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; and 9 more (refs \#2762, \#2875, \#2750, \#2897, \#3242, …).

crates/rig-core/src/completion/history · medium confidence · unverified

12 commits (6 fixes) modifying plugins/rig-steel

A change to existing behaviour in plugins/rig-steel — 12 commits (6 fixs), 6 files: “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “fix!: every limit on asked work is a settings.toml key with a sensible default, and rig says what it changes (\#2995)”; and 9 more (refs \#2984, \#3202, \#2995, \#2848, \#2985, …).

plugins/rig-steel · medium confidence · unverified

13 commits (5 fixes) modifying corpus/shared

A change to existing behaviour in corpus/shared — 13 commits (5 fixs), 15 files: “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “fix!: every limit on asked work is a settings.toml key with a sensible default, and rig says what it changes (\#2995)”; “refactor(rig-tools)!: one listing rule for both watch backends, and typed watch events (\#3270)”; and 10 more (refs \#3202, \#2995, \#3270, \#2838, \#2898, …).

corpus/shared · medium confidence · unverified

15 commits (4 fixes) modifying crates/rig-cassette/src/agent/replay

A change to existing behaviour in crates/rig-cassette/src/agent/replay — 15 commits (4 fixs), 1 file: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “fix(agent): a 64k output budget, unknown-tool feedback, honest failed summaries, ordered headless JSON (\#2894)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; and 12 more (refs \#2640, \#2894, \#2750, \#3003, \#2617, …).

crates/rig-cassette/src/agent/replay · medium confidence · unverified

17 commits (3 fixes) modifying test-support

A change to existing behaviour in test-support — 17 commits (3 fixs), 4 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “feat(acp): rig acp, an ACP v1 front with a rig process per session, on rig-mcp-client (\#2948)”; “refactor: delete the reflect and replay features and the source-shape guards (\#2523)”; and 14 more (refs \#2973, \#2948, \#2523, \#2665, \#2994, …).

test-support · medium confidence · unverified

2 commits (1 fix) modifying crates/rig-agent/src/agent/telemetry

A change to existing behaviour in crates/rig-agent/src/agent/telemetry — 2 commits (1 fix), 1 file: “fix(core)!: stop forcing serde\_json features (\#2934)”; “feat(telemetry): record tool definitions on agent completion spans (\#2885)” (refs \#2934, \#2885).

crates/rig-agent/src/agent/telemetry · medium confidence · unverified

20 commits (10 fixes) modifying plugins/rig-claude-cloak

A change to existing behaviour in plugins/rig-claude-cloak — 20 commits (10 fixs), 214 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “refactor(rig-claude-cloak): credentials are kept through rig-core's FileTokenStore (\#3305)”; and 17 more (refs \#2973, \#3202, \#3305, \#2897, \#3293, …).

plugins/rig-claude-cloak · medium confidence · unverified

20 commits (6 fixes) modifying crates/rig-core/src/completion/message

A change to existing behaviour in crates/rig-core/src/completion/message — 20 commits (6 fixs), 6 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(rig-agent)!: runs survive a restart at any point: one answer owner, resumable in-flight steps, executable-only CallTools (\#2762)”; “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; and 17 more (refs \#2918, \#2762, \#2640, \#2433, \#3047, …).

crates/rig-core/src/completion/message · medium confidence · unverified

21 commits (6 fixes) modifying crates/rig-typesafeai

A change to existing behaviour in crates/rig-typesafeai — 21 commits (6 fixs), 14 files: “feat(typesafeai): set a loaded JevConfig's token (\#3180)”; “feat: initial typesafeai provider (jev) (\#2550)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; and 18 more (refs \#3180, \#2550, \#2750, \#3118, \#2845, …).

crates/rig-typesafeai · medium confidence · unverified

22 commits (11 fixes) modifying crates/rig-core/src/catalog

A change to existing behaviour in crates/rig-core/src/catalog — 22 commits (11 fixs), 11 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “feat(harness): add claude-cloak and harden Anthropic handling (\#2875)”; “fix(catalog, compaction): fit models' input limits with simpler overflow recovery (\#2880)”; and 19 more (refs \#2973, \#2875, \#2880, \#2915, \#2750, …).

crates/rig-core/src/catalog · medium confidence · unverified

22 commits (2 fixes) modifying crates/rig-rmcp

A change to existing behaviour in crates/rig-rmcp — 22 commits (2 fixs), 11 files: “chore: move every inline test module to a sibling file (\#2433)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “feat!: lower ModelHandle and the erased tool set into rig-core; pure rig\_run::prepare\_request (\#2405)”; and 19 more (refs \#2433, \#3202, \#2405, \#2643, \#2575, …).

crates/rig-rmcp · medium confidence · unverified

22 commits (5 fixes) modifying crates/rig-cassette/src/http

A change to existing behaviour in crates/rig-cassette/src/http — 22 commits (5 fixs), 19 files: “docs: simplify production comments and preserve contract rationale (\#2568)”; “feat(core)!: stateful handle round-trips and reasoning provenance (\#2578)”; “feat(gemini): automatic explicit caching for long runs (\#2628)”; and 19 more (refs \#2568, \#2578, \#2628, \#2974, \#2761, …).

crates/rig-cassette/src/http · medium confidence · unverified

23 commits (4 fixes) modifying crates/rig-core/src/effect

A change to existing behaviour in crates/rig-core/src/effect — 23 commits (4 fixs), 3 files: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “fix(agent): a 64k output budget, unknown-tool feedback, honest failed summaries, ordered headless JSON (\#2894)”; “chore!: remove the rig-ecs crate and its tests, goldens and fixtures (\#2761)”; and 20 more (refs \#2640, \#2894, \#2761, \#2617, \#2535, …).

crates/rig-core/src/effect · medium confidence · unverified

23 commits (5 fixes) modifying crates/rig-core/src/operation/completion

A change to existing behaviour in crates/rig-core/src/operation/completion — 23 commits (5 fixs), 1 file: “feat(harness): add claude-cloak and harden Anthropic handling (\#2875)”; “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “feat(core)!: stateful handle round-trips and reasoning provenance (\#2578)”; and 20 more (refs \#2875, \#2640, \#2578, \#3285, \#2750, …).

crates/rig-core/src/operation/completion · medium confidence · unverified

24 commits (5 fixes) modifying crates/rig-cassette/src/effect\_log

A change to existing behaviour in crates/rig-cassette/src/effect\_log — 24 commits (5 fixs), 14 files: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “docs: simplify production comments and preserve contract rationale (\#2568)”; “fix(agent): a 64k output budget, unknown-tool feedback, honest failed summaries, ordered headless JSON (\#2894)”; and 21 more (refs \#2640, \#2568, \#2894, \#2750, \#2845, …).

_crates/rig-cassette/src/effect\log · medium confidence · unverified

26 commits (5 fixes) modifying crates/rig-core/src/error

A change to existing behaviour in crates/rig-core/src/error — 26 commits (5 fixs), 6 files: “fix(catalog, compaction): fit models' input limits with simpler overflow recovery (\#2880)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; “feat(rig-core): RateLimits::from\_headers reads a reply's request and token allowances (\#3304)”; and 23 more (refs \#2880, \#2750, \#3304, \#3287, \#2897, …).

crates/rig-core/src/error · medium confidence · unverified

27 commits (3 fixes) modifying crates/rig-reqwest

A change to existing behaviour in crates/rig-reqwest — 27 commits (3 fixs), 16 files: “feat(harness): add claude-cloak and harden Anthropic handling (\#2875)”; “chore: move every inline test module to a sibling file (\#2433)”; “refactor!: rig-reqwest API and hygiene cleanups (\#2428)”; and 24 more (refs \#2875, \#2433, \#2428, \#2568, \#3202, …).

crates/rig-reqwest · medium confidence · unverified

27 commits (6 fixes) modifying crates/rig-core/src/completion/request

A change to existing behaviour in crates/rig-core/src/completion/request — 27 commits (6 fixs), 3 files: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “chore: move every inline test module to a sibling file (\#2433)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 24 more (refs \#2640, \#2433, \#3202, \#2880, \#2750, …).

crates/rig-core/src/completion/request · medium confidence · unverified

27 commits (7 fixes) modifying crates/rig-core/src/driver

A change to existing behaviour in crates/rig-core/src/driver — 27 commits (7 fixs), 5 files: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “fix(catalog, compaction): fit models' input limits with simpler overflow recovery (\#2880)”; “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; and 24 more (refs \#2640, \#2880, \#2750, \#2974, \#2897, …).

crates/rig-core/src/driver · medium confidence · unverified

28 commits (2 fixes) modifying crates/rig-vectorize

A change to existing behaviour in crates/rig-vectorize — 28 commits (2 fixs), 11 files: “refactor: workspace-wide LOC consolidation pass 7 (net −366 production lines) (\#2310)”; “chore: move every inline test module to a sibling file (\#2433)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 25 more (refs \#2310, \#2433, \#3202, \#2545, \#1714, …).

crates/rig-vectorize · medium confidence · unverified

3 commits (0 fixes) modifying crates/rig-core/src/completion/prompt

A change to existing behaviour in crates/rig-core/src/completion/prompt — 3 commits, 1 file: “perf(rig)!: measured performance pass on rig's libraries, with benchmarks (\#2796)”; “docs(rig-core, rig-memory): base-crate docs stop naming the agent's crates (\#3301)”; “feat(rig-harness): a thin coding agent over rig libraries, with rig-ecs and rig-tools (\#2771)” (refs \#2796, \#3301, \#2771).

crates/rig-core/src/completion/prompt · medium confidence · unverified

3 commits (1 fix) modifying crates/rig-core/src/completion/cache\_cost

A change to existing behaviour in crates/rig-core/src/completion/cache\_cost — 3 commits (1 fix), 1 file: “fix(usage)!: one meaning for Usage on every provider (\#2631)”; “test(cache): long-run cache proof and one cache cost report across providers (\#2629)”; “test: shape-matched cassettes, encoder snapshots, and a coverage gate (\#2720)” (refs \#2631, \#2629, \#2720).

_crates/rig-core/src/completion/cache\cost · medium confidence · unverified

33 commits (14 fixes) modifying corpus/sp

A change to existing behaviour in corpus/sp — 33 commits (14 fixs), 34 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 30 more (refs \#2973, \#2984, \#3202, \#3270, \#3048, …).

corpus/sp · medium confidence · unverified

33 commits (15 fixes) modifying corpus/ho

A change to existing behaviour in corpus/ho — 33 commits (15 fixs), 65 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 30 more (refs \#2973, \#2984, \#3202, \#3270, \#2852, …).

corpus/ho · medium confidence · unverified

34 commits (9 fixes) modifying crates/rig-core/src/serve

A change to existing behaviour in crates/rig-core/src/serve — 34 commits (9 fixs), 12 files: “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “feat(core)!: stateful handle round-trips and reasoning provenance (\#2578)”; “fix(agent): a 64k output budget, unknown-tool feedback, honest failed summaries, ordered headless JSON (\#2894)”; and 31 more (refs \#3202, \#2578, \#2894, \#2845, \#2502, …).

crates/rig-core/src/serve · medium confidence · unverified

37 commits (16 fixes) modifying corpus/sn

A change to existing behaviour in corpus/sn — 37 commits (16 fixs), 68 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; and 34 more (refs \#2918, \#2973, \#2984, \#3202, \#2880, …).

corpus/sn · medium confidence · unverified

37 commits (18 fixes) modifying corpus/fh

A change to existing behaviour in corpus/fh — 37 commits (18 fixs), 62 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 34 more (refs \#2973, \#2984, \#3202, \#3290, \#2995, …).

corpus/fh · medium confidence · unverified

38 commits (16 fixes) modifying corpus/oc

A change to existing behaviour in corpus/oc — 38 commits (16 fixs), 48 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 35 more (refs \#2973, \#2984, \#3202, \#3290, \#2880, …).

corpus/oc · medium confidence · unverified

39 commits (9 fixes) modifying crates/rig-core/src/operation

A change to existing behaviour in crates/rig-core/src/operation — 39 commits (9 fixs), 4 files: “feat(harness): add claude-cloak and harden Anthropic handling (\#2875)”; “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “feat(core)!: stateful handle round-trips and reasoning provenance (\#2578)”; and 36 more (refs \#2875, \#2640, \#2578, \#2729, \#3285, …).

crates/rig-core/src/operation · medium confidence · unverified

4 commits (0 fixes) modifying plugins/rig-traffic

A change to existing behaviour in plugins/rig-traffic — 4 commits, 4 files: “chore: release v0.46.0 (\#2958)”; “feat(rig-traffic): record every provider HTTP exchange, on rig-cassette's engine (\#2893)”; “feat(harness,cli,mcp-client)!: one settings file, plugins from the home list, full trust in projects, and the launcher in its own crate (\#3005)”; and 1 more (refs \#2958, \#2893, \#3005, \#3268).

plugins/rig-traffic · medium confidence · unverified

4 commits (1 fix) modifying crates/rig-core/src/test\_utils/raw\_parity

A change to existing behaviour in crates/rig-core/src/test\_utils/raw\_parity — 4 commits (1 fix), 1 file: “feat: typed generation options, model catalog, provider extensions and normalized replies (\#2750)”; “chore!: remove the rig-ecs crate and its tests, goldens and fixtures (\#2761)”; “fix(core)!: stop forcing serde\_json features (\#2934)”; and 1 more (refs \#2750, \#2761, \#2934, \#2835).

_crates/rig-core/src/test\_utils/raw\parity · medium confidence · unverified

4 commits (2 fixes) modifying crates/rig-core/src/catalog/connector

A change to existing behaviour in crates/rig-core/src/catalog/connector — 4 commits (2 fixs), 1 file: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “perf(rig)!: measured performance pass on rig's libraries, with benchmarks (\#2796)”; “fix(models)!: /model lists only models you can use now; a model that doesn't connect is refused and the current one stays (\#2983)”; and 1 more (refs \#2973, \#2796, \#2983, \#2771).

crates/rig-core/src/catalog/connector · medium confidence · unverified

40 commits (20 fixes) modifying corpus/base

A change to existing behaviour in corpus/base — 40 commits (20 fixs), 112 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; and 37 more (refs \#2918, \#2973, \#2984, \#2873, \#3202, …).

corpus/base · medium confidence · unverified

44 commits (6 fixes) modifying crates/rig-core/src/embeddings

A change to existing behaviour in crates/rig-core/src/embeddings — 44 commits (6 fixs), 7 files: “chore: move every inline test module to a sibling file (\#2433)”; “feat: carry the provider transport request id on completion errors (\#2314) (\#2315)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 41 more (refs \#2433, \#2314, \#2315, \#3202, \#2389, …).

crates/rig-core/src/embeddings · medium confidence · unverified

44 commits (7 fixes) modifying crates/rig-core/src/vector\_store

A change to existing behaviour in crates/rig-core/src/vector\_store — 44 commits (7 fixs), 8 files: “refactor: workspace-wide LOC consolidation pass 7 (net −366 production lines) (\#2310)”; “chore: move every inline test module to a sibling file (\#2433)”; “refactor: workspace-wide LOC consolidation pass 8 (net −1,353 production lines) (\#2320)”; and 41 more (refs \#2310, \#2433, \#2320, \#2132, \#2645, …).

_crates/rig-core/src/vector\store · medium confidence · unverified

45 commits (6 fixes) modifying crates/rig-sqlite

A change to existing behaviour in crates/rig-sqlite — 45 commits (6 fixs), 10 files: “fix(sqlite): deserialize JSON columns (\#1797)”; “chore: move every inline test module to a sibling file (\#2433)”; “chore: release v0.39.0 (\#1888)”; and 42 more (refs \#1797, \#2433, \#1888, \#3202, \#2389, …).

crates/rig-sqlite · medium confidence · unverified

5 commits (0 fixes) modifying examples/agent\_tool\_call\_streaming

A change to existing behaviour in examples/agent\_tool\_call\_streaming — 5 commits, 4 files: “feat(rig-agent): a streamed run reports each tool's accepted output as it finishes (\#3314)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “feat(rig-agent): a streamed run reports each tool call as it starts (\#3313)”; and 2 more (refs \#3314, \#3202, \#3313, \#2721, \#2835).

_examples/agent\_tool\_call\streaming · medium confidence · unverified

56 commits (14 fixes) modifying crates/rig-core/src/tool

A change to existing behaviour in crates/rig-core/src/tool — 56 commits (14 fixs), 25 files: “refactor(rig-core, rig-ecs)!: the kernel says how much of a call's facts a session keeps (\#3283)”; “refactor!: assistant content is tagged and provider extras are a named field (\#2277)”; “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; and 53 more (refs \#3283, \#2277, \#2640, \#2433, \#2055, …).

crates/rig-core/src/tool · medium confidence · unverified

6 commits (1 fix) modifying crates/rig-cli

A change to existing behaviour in crates/rig-cli — 6 commits (1 fix), 17 files: “fix(rig-cli): the version names its commit, the lock follows the checkout, and a source mismatch is refused (\#3221)”; “feat(rig-cli): rig plugin list groups rig's plugins and yours; each build writes a writing-rig-plugins skill (\#3218)”; “feat(rig-cli): warn once per start of a 0.46 home's plugins.toml and defaults.json (\#3239)”; and 3 more (refs \#3221, \#3218, \#3239, \#3245, \#3217, …). Also in this area: 5 commits (3 fixes) fixing crates/rig-cli.

crates/rig-cli, crates/rig-markdown · medium confidence · unverified

6 commits (3 fixes) modifying crates/rig-ecs

A change to existing behaviour in crates/rig-ecs — 6 commits (3 fixs), 7 files: “refactor(rig-core, rig-ecs)!: the kernel says how much of a call's facts a session keeps (\#3283)”; “fix(rig-core): a wait hint with minutes or hours ("6m0s") is read (\#3287)”; “fix(rig-core, rig-ecs, rig-basics)!: one reader and printer for a reasoning setting as text, so \[model\] effort = "high" works on budget-only models (\#3291)”; and 3 more (refs \#3283, \#3287, \#3291, \#3303, \#3277, …).

crates/rig-ecs/tests · medium confidence · unverified

7 commits (0 fixes) modifying crates/rig-core/src/model/listing

A change to existing behaviour in crates/rig-core/src/model/listing — 7 commits, 1 file: “chore: move every inline test module to a sibling file (\#2433)”; “feat(copilot)!: expose whether a listed model supports tool calls (\#3116)”; “refactor!: archimpro — one typed decoder and one fold per wire (\#2617)”; and 4 more (refs \#2433, \#3116, \#2617, \#2611, \#2538, …).

crates/rig-core/src/model/listing · medium confidence · unverified

7 commits (2 fixes) modifying test-support/rig-harness-test-support

A change to existing behaviour in test-support/rig-harness-test-support — 7 commits (2 fixs), 1 file: “chore: drop build edges nothing uses (\#3144)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “ci!: PRs run what the change can break, one test binary per crate, one required check (\#3014)”; and 4 more (refs \#3144, \#3202, \#3014, \#2963, \#2771, …).

test-support/rig-harness-test-support · medium confidence · unverified

8 commits (0 fixes) modifying examples/agent\_with\_retry\_hook

A change to existing behaviour in examples/agent\_with\_retry\_hook — 8 commits, 1 file: “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “refactor(client)!: single canonical CompletionClient + AgentClientExt (\#2205)”; “refactor!: provider clients own their transport; conversations use plain values (\#2611)”; and 5 more (refs \#3202, \#2205, \#2611, \#2538, \#2182, …).

_examples/agent\_with\_retry\hook · medium confidence · unverified

8 commits (2 fixes) modifying crates/rig-core/src/wire/secret

A change to existing behaviour in crates/rig-core/src/wire/secret — 8 commits (2 fixs), 1 file: “refactor!: replace NonEmpty\<T\> with Vec\<T\>, check empty turns at the request boundary (\#2640)”; “refactor!: archimpro — one typed decoder and one fold per wire (\#2617)”; “refactor!: provider clients own their transport; conversations use plain values (\#2611)”; and 5 more (refs \#2640, \#2617, \#2611, \#3111, \#2538, …).

crates/rig-core/src/wire/secret · medium confidence · unverified

9 commits (2 fixes) modifying corpus/library

A change to existing behaviour in corpus/library — 9 commits (2 fixs), 13 files: “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “feat(tools,basics)!: remove the unused move tool and the checkpoint plugin (\#3048)”; “refactor(rig-harness-protocol)!: file errors name the path and step; only NotFound is absent (\#3275)”; and 6 more (refs \#3202, \#3048, \#3275, \#2848, \#2999, …).

corpus/library · medium confidence · unverified

9 commits (2 fixes) modifying crates/rig-core/src/loaders/epub

A change to existing behaviour in crates/rig-core/src/loaders/epub — 9 commits (2 fixs), 4 files: “chore: move every inline test module to a sibling file (\#2433)”; “docs: set doctest to true, and update doc comments (\#1716)”; “docs: finish the comment cleanup across the workspace (\#2575)”; and 6 more (refs \#2433, \#1716, \#2575, \#2409, \#2305, …).

crates/rig-core/src/loaders/epub · medium confidence · unverified

Agent construction and execution refactored around a bus-driven architecture

The agent builder now constructs agents over an owned or host-driven bus, enabling typestate enforcement for tool configuration (builder-supplied tools versus a shared tool server) and scoped model registrations. The agent's lifecycle is driven by a shared bus with a per-poll driver lock and a bus-wide waker set, ensuring that multiple runs on the same bus coordinate correctly without deadlocks. Request preparation is centralized, applying hook patches (preamble, temperature, tool choice, context) and dynamic tool retrieval before dispatch. Recording is refused on host buses to prevent silent failures, and a warning is emitted when a conversation is created without a memory backend. Also in this area: Agent run resilience and streaming tool item projection; Agent engine ensures one-to-one completion dispatch/outcome and per-turn tool choice policy; Bedrock and Candle providers now support typed generation options and provider extensions; Agent run configuration and API updates.

crates/rig-agent/src/agent · high confidence

Agent engine enforces single-owner dispatch and projects tool results from committed history

The agent engine now guarantees that every dispatched effect (model completion or tool call) has exactly one owner responsible for minting the ID, firing the dispatch event, and closing it with the outcome, ensuring that in-flight steps are resumable and runs survive restarts. Model completions are strictly limited to a single dispatch per scope, preventing duplicate sends, while tool stream items are now generated as a projection of committed history, ensuring that execution commits and results are correctly paired and ordered in the output stream.

crates/rig-agent/src/agent/engine · high confidence

Agent run state machine refactored for resumability and structured tool handling

The agent run logic in \rig-agent\ has been restructured into a serializable, sans-I/O state machine that supports resuming runs from any point. Tool calls are now managed through a \ToolBatch\ system that classifies calls as executable, malformed, or pre-answered, enabling drivers to handle invalid tool arguments via hooks and retry policies. The run maintains a \CommittedLog\ of messages, allowing drivers to stream tool results and history projections without re-processing committed data. Per-turn request preparation is now handled by \RequestPatch\, which allows hooks to override parameters like temperature, tool choice, and context documents without mutating the agent's baseline configuration. The run's state, including tool policies and committed history, is fully serializable, enabling state round-trips for persistence and recovery.

crates/rig-agent/src/run · high confidence

Hook stack now chains rewrites in registration order

The agent's hook system now processes tool call and result rewrites in the order they are registered. When multiple hooks modify tool arguments or results, each subsequent hook sees the output of the previous one, allowing for cumulative transformations (e.g., step 0 -\> step 1 -\> step 2) rather than independent or unordered modifications.

crates/rig-agent/src/agent/hook · high confidence

Migrate RAG examples to the new Tool API and package layout

The RAG dynamic tools examples have been updated to use the revised Tool trait, replacing the old \definition\ method with \parameters\ and \description\, and updating the \call\ signature to accept typed arguments and return typed results. Additionally, the examples have been reorganized into a package-per-example directory structure, and the calculator chatbot example now uses the new \ChatBotBuilder\ integration.

_examples/rag\_dynamic\tools · high confidence

Refined \#\[derive(Embed)\] macro with stricter attribute validation

The \\#\[derive(Embed)\]\ macro in \crates/rig-derive/src/embed\ has been updated to enforce stricter rules on field annotations. It now explicitly rejects fields that mix basic \\#\[embed\]\ tags with custom \\#\[embed(embed\_with = "...")\]\ attributes, and prevents duplicate \embed\_with\ attributes on a single field. Additionally, the macro now validates that \embed\_with\ values are valid string literals without unexpected suffixes, ensuring more robust compile-time errors for users defining embedding logic.

crates/rig-derive/src/embed · high confidence

Vertex AI response normalization and decoding

The Vertex AI provider now normalizes completion responses by translating raw SDK data into a standard REST-JSON format. This process ensures that enum values (such as finish reasons, safety ratings, and modality types) are represented by their string names rather than numeric codes, and it correctly aggregates tool-use token counts. The provider implements a dedicated decoder and document reassembly logic to handle these normalized responses consistently with other providers. Also in this area: Reasoning content blobs are now captured for replay; Vertex AI completion error classification and image tool result encoding; Refined error classification and removed ECS-specific examples; Bedrock and Candle providers now support typed generation options and provider extensions; Added Vertex AI integration tests for history conformance, raw API, and runtime ownership; and 4 more.

crates/rig-vertexai · high confidence

Fixes

1 commit (1 fix) fixing plugins/rig-fold

A fix in plugins/rig-fold — 1 commit (1 fix), 2 files: “fix(rig-fold): draw a result whole unless folding it saves two rows (\#3251)” (refs \#3251). Also in this area: 1 commit (0 fixes) modifying corpus/fh, plugins/rig-fold.

plugins/rig-fold · low confidence · unverified

12 commits (7 fixes) fixing crates/rig-core/src/conversation

A fix in crates/rig-core/src/conversation — 12 commits (7 fixs), 10 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(ecs)!: a stopped turn keeps what streamed, and a stop is never an error (\#2966)”; “fix(ecs)!: a call is known by its reply and place; every stop goes through one owner (\#2985)”; and 9 more (refs \#2918, \#2966, \#2985, \#2796, \#2895, …).

crates/rig-core/src/conversation · medium confidence · unverified

2 commits (2 fixes) fixing corpus/rf-01

A fix in corpus/rf-01 — 2 commits (2 fixs), 3 files: “fix!: everything the pre-0.45 API review found (\#2848)”; “fix(basics): settle a helper's count with try\_update, which Rust 1.99 does not deprecate (\#3215)” (refs \#2848, \#3215).

corpus/rf-01 · medium confidence · unverified

20 commits (12 fixes) fixing plugins/rig-mcp-client

A fix in plugins/rig-mcp-client — 20 commits (12 fixs), 9 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “fix!: every limit on asked work is a settings.toml key with a sensible default, and rig says what it changes (\#2995)”; and 17 more (refs \#2973, \#3202, \#2995, \#2948, \#3133, …).

plugins/rig-mcp-client · medium confidence · unverified

28 commits (16 fixes) fixing plugins/rig-acp

A fix in plugins/rig-acp — 28 commits (16 fixs), 19 files: “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “feat(rig-ecs)!: write a reply that did not stream as stream deltas (\#3264)”; “feat(acp): rig acp, an ACP v1 front with a rig process per session, on rig-mcp-client (\#2948)”; and 25 more (refs \#2973, \#3264, \#2948, \#3221, \#3048, …).

plugins/rig-acp · medium confidence · unverified

3 commits (3 fixes) fixing crates/rig-core/src/catalog/read

A fix in crates/rig-core/src/catalog/read — 3 commits (3 fixs), 1 file: “fix(catalog, compaction): fit models' input limits with simpler overflow recovery (\#2880)”; “fix(core)!: stop forcing serde\_json features (\#2934)”; “fix(catalog)!: overrides reach every wire, one lookup and reference rule, connect from the environment (\#2769)” (refs \#2880, \#2934, \#2769).

crates/rig-core/src/catalog/read · medium confidence · unverified

35 commits (18 fixes) fixing corpus/ct

A fix in corpus/ct — 35 commits (18 fixs), 63 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; and 32 more (refs \#2918, \#2973, \#2984, \#3202, \#3290, …).

corpus/ct · medium confidence · unverified

4 commits (3 fixes) fixing corpus/rf-03

A fix in corpus/rf-03 — 4 commits (3 fixs), 3 files: “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; “fix(agent)!: quitting mid-turn is a stop, and a restart carries nothing on by itself (\#2943)”; “fix!: everything the pre-0.45 API review found (\#2848)”; and 1 more (refs \#3202, \#2943, \#2848, \#2991).

corpus/rf-03 · medium confidence · unverified

4 commits (3 fixes) fixing crates/rig-cassette/src/scrub

A fix in crates/rig-cassette/src/scrub — 4 commits (3 fixs), 1 file: “fix(cassette,xtask,test-support): scrub team ids, price one-hour cache writes, close the guard's and registry's gaps (\#3045)”; “fix(io)!: input and output are bounded; a stop is never blocked by output (\#2974)”; “fix(chatgpt, tui): ask ChatGPT for a reasoning summary, draw reasoning as markdown (\#2912)”; and 1 more (refs \#3045, \#2974, \#2912, \#2893).

crates/rig-cassette/src/scrub · medium confidence · unverified

40 commits (21 fixes) fixing crates/rig-harness-protocol/src

A fix in crates/rig-harness-protocol/src — 40 commits (21 fixs), 11 files: “fix(harness): an existing private file's staged copy is never readable by others (\#3166)”; “fix(harness): create session directories private to the user (\#3137)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 37 more (refs \#3166, \#3137, \#3202, \#2995, \#2948, …).

crates/rig-harness-protocol/src · medium confidence · unverified

47 commits (28 fixes) fixing crates/rig-tools

A fix in crates/rig-tools — 47 commits (28 fixs), 38 files: “fix(rig-tools): triobench round 6: shell output drops terminal escape codes, a cut keeps the first and last errors, a remote write cut off mid-way says whether it happened, an edit after a write is one call (\#2858)”; “fix(harness): an existing private file's staged copy is never readable by others (\#3166)”; “refactor!: remove duplicated tests, dead code and doc bloat across the workspace (\#3202)”; and 44 more (refs \#2858, \#3166, \#3202, \#2995, \#2913, …).

crates/rig-tools · medium confidence · unverified

5 commits (3 fixes) fixing crates/rig-cli

A fix in crates/rig-cli — 5 commits (3 fixs), 17 files: “fix(rig-cli): a signal at a secret prompt puts the terminal back before the launcher ends (\#3281)”; “refactor(rig-harness-protocol)!: file errors name the path and step; only NotFound is absent (\#3275)”; “fix(rig-cli): a staged file that is not a program is passed over on macOS too (\#3307)”; and 2 more (refs \#3281, \#3275, \#3307, \#3276, \#3297). Also in this area: 6 commits (1 fix) modifying crates/rig-cli.

crates/rig-cli/src · medium confidence · unverified

50 commits (26 fixes) fixing plugins/rig-tui

A fix in plugins/rig-tui — 50 commits (26 fixs), 46 files: “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; “feat(rig-basics): a welcome at the top of the transcript: rig's mark, version, model and directory (\#2861)”; and 47 more (refs \#2918, \#2973, \#2861, \#2873, \#3202, …).

plugins/rig-tui · medium confidence · unverified

54 commits (29 fixes) fixing crates/rig-harness/src

A fix in crates/rig-harness/src — 54 commits (29 fixs), 51 files: “fix(rig-tools): triobench round 6: shell output drops terminal escape codes, a cut keeps the first and last errors, a remote write cut off mid-way says whether it happened, an edit after a write is one call (\#2858)”; “fix(tools)!: --tools limits only the run it is given to (\#2984)”; “feat(rig-basics)!: AGENTS.md and skills as plugins, /login with API keys, Ctrl+D twice quits (\#2873)”; and 51 more (refs \#2858, \#2984, \#2873, \#3137, \#3202, …).

crates/rig-harness/src · medium confidence · unverified

55 commits (31 fixes) fixing crates/rig-ecs/src

A fix in crates/rig-ecs/src — 55 commits (31 fixs), 25 files: “refactor(rig-core, rig-ecs)!: the kernel says how much of a call's facts a session keeps (\#3283)”; “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; and 52 more (refs \#3283, \#2918, \#2973, \#2984, \#3264, …).

crates/rig-ecs/src · medium confidence · unverified

99 commits (59 fixes) fixing plugins/rig-basics

A fix in plugins/rig-basics — 99 commits (59 fixs), 85 files: “fix(rig-tools): triobench round 6: shell output drops terminal escape codes, a cut keeps the first and last errors, a remote write cut off mid-way says whether it happened, an edit after a write is one call (\#2858)”; “fix(agent)!: Esc keeps the partial reply, /compact keeps history, accurate summaries, ctx ? after compaction (\#2918)”; “fix(models)!: only an explicit choice becomes the default; /model offers only models that answer (\#2973)”; and 96 more (refs \#2858, \#2918, \#2973, \#2875, \#2984, …).

plugins/rig-basics · medium confidence · unverified

Gemini gRPC client rejects invalid API keys before connecting

The \rig-gemini-grpc\ client now validates the API key during construction, rejecting keys containing newlines (or other invalid characters) with an \InvalidApiKey\ error before attempting to establish a gRPC connection. This prevents wasted connection attempts and provides immediate feedback on configuration errors.

crates/rig-gemini-grpc · high confidence

PostgreSQL vector store search query generation and filtering fixes

The \rig-postgres\ crate introduces a new PostgreSQL vector store implementation that corrects several SQL generation bugs in search queries. Specifically, it fixes parameter placeholder consistency (ensuring all operators use \$\ instead of mixing \?\), corrects the \IN\ clause syntax for member filters, and fixes threshold filtering to properly calculate minimum similarity per distance function rather than comparing raw distances. The implementation also ensures correct parameter binding order when combining thresholds with compound filters.

crates/rig-postgres/src · high confidence

Test coverage

1 commit adding/updating tests in crates/rig-core/tests/fixtures; 1 commit adding/updating tests in crates/rig-derive/tests/integration, crates/rig-gemini-grpc/tests; 11 commits adding/updating tests in crates/rig-cassette/tests/providers/chatgpt; 2 commits adding/updating tests in crates/rig-core/tests/integration, crates/rig-memory/tests; 2 commits adding/updating tests in crates/rig-derive/tests/ui/custom\_name; 3 commits adding/updating tests in crates/rig-agent/tests; 3 commits adding/updating tests in crates/rig-candle/tests; 3 commits adding/updating tests in crates/rig-core/src/audio\_generation, crates/rig-core/src/image\_generation, crates/rig-core/src/transcription; 3 commits adding/updating tests in crates/rig-core/src/completion/history/tests; 4 commits adding/updating tests in crates/rig-core/tests; 5 commits adding/updating tests in crates/rig-cli/tests; 6 commits adding/updating tests in crates/rig-core/tests/common; 7 commits adding/updating tests in crates/rig-cassette/tests/common/corpus\_matrix; Added Anthropic cassette tests for agent effects, citations, and hooks; Added Bedrock cassette replay tests for agent, embeddings, and raw data capture; Added OpenRouter provider tests for audio generation and document handling; Added UI test for invalid handler registration; Added UI tests for ToolContext validation rules; Added UI tests for derive macro validation errors; Added agent run stepper fixture for testing core integration; Added benchmark suite for rig-cassette agent and wire performance; Added cassette-backed test coverage for the Venice provider; Added compile-time tests for renamed dependency resolution in rig-derive; Added comprehensive cassette tests for the llama.cpp provider; Added comprehensive test coverage for the Mistral provider; Added comprehensive test infrastructure for rig-cassette; Added comprehensive tests for Gemini gRPC completion and transport layer; Added corpus-based golden replay tests for rig-cassette; Added golden-request tests for agent completion preparation; Added history conformance tests for the Bedrock Converse wire; Added integration tests for the Copilot provider; Added live smoke tests for OpenAI audio, image, document, and websocket capabilities; Added model conformance tests for tool usage and protocol hygiene; Added runtime tests for rig-agent's agent loop, tool lifecycle, and turn termination; Added test fixture to verify tool facade re-exports and derive macro context detection; Added test for Qdrant filter availability in search requests; Added test suite for Anthropic provider cassettes; Added tests for agent builder configuration and recording constraints; Added tests for agent memory persistence and error handling; Added tests for extractor dynamic context and usage accumulation; Added tests for typed generation options across multiple providers; Added unit tests for Gemini gRPC generation options and request encoding; Added vector search benchmark suite; Added xAI provider test suite; Agent runner test suite reorganized into dedicated modules; Comprehensive test coverage for rig-cassette cassettes; Expanded Anthropic cassette test coverage for agent effects and reasoning; Expanded OpenRouter cassette test coverage for agent, tool, and multimodal workflows; Expanded test coverage for Gemini and llama.cpp provider cassettes; Expanded test coverage for core agent behaviors and system invariants; New history conformance test suite for completion wires; New test harnesses for history conformance and long-run cache accounting; New test-support crate for validating default plugin configuration; Robust handling of malformed streamed tool arguments; Tests added for provider error header and ID forwarding.

Dependencies

Rig workspace restructured into versioned companion crates

The Rig library has been split into a modular workspace of versioned companion crates (rig-core, rig-agent, rig-derive, rig-http, rig-reqwest, rig-tungstenite, rig-cassette, rig-ecs, rig-memory, and various provider/vector-store integrations), all unified under a facade crate and pinned to version 0.42.0. This reorganization decouples the core runtime from specific transports and providers, allowing users to depend only on the components they need (e.g., rig-core without the bundled reqwest transport) and enabling independent versioning of integrations like AWS Bedrock, Google Gemini, and vector stores such as LanceDB, MongoDB, and Qdrant.

(dependencies) · high confidence

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

How this codebase got here

Score

  • CAI 66 → 72 (+6.3)
  • Rubric changed (rubric-2026.10.5 → rubric-2026.11.4) — scores are not directly comparable.

Lenses

  • Code Health 76 → 73 (-2.7)
  • Architecture 96 → 95 (-0.2)
  • Maturity 65 → 69 (+4.4)
  • Readiness 58 → 69 (+11.0)
  • Security 71 → 75 (+3.3)
  • Domain Modelling 91 → 93 (+2.1)
  • Event Sourcing 100 → 100 (+0.0)
  • Performance 97 → 97 (+0.0)

Resolved (180)

  • Change coupling: streaming.rs ↔ streaming.rs (crates/rig-bedrock/src/streaming.rs)
  • Change coupling: streaming.rs ↔ streaming.rs (crates/rig-bedrock/src/streaming.rs)
  • Duplicated block (10 lines × 2) (corpus/base/src/check_after_turn.rs)
  • Duplicated block (10 lines × 2) (corpus/oc/src/dataset_cli.rs)
  • Duplicated block (10 lines × 2) (crates/rig-core/src/providers/openai/options.rs)
  • Duplicated block (10 lines × 2) (crates/rig-milvus/src/lib.rs)
  • Duplicated block (11 lines × 2) (crates/rig-core/src/providers/gemini/image_generation.rs)
  • Duplicated block (12 lines × 2) (corpus/base/src/custom_provider.rs)
  • Duplicated block (13 lines × 2) (corpus/oc/src/dataset_cli.rs)
  • Duplicated block (13 lines × 2) (crates/rig-core/src/providers/openai/options.rs)
  • Duplicated block (2–5 lines × 3) (crates/rig-core/src/providers/openai/options.rs)
  • Duplicated block (5 lines × 2) (corpus/base/src/mcp_client.rs)
  • Duplicated block (5 lines × 2) (corpus/base/src/skills.rs)
  • Duplicated block (5 lines × 2) (corpus/oc/src/dataset_cli/dataset.rs)
  • Duplicated block (5 lines × 2) (crates/rig-core/src/providers/gemini/extension.rs)
  • Duplicated block (5 lines × 2) (crates/rig-tools/src/edit.rs)
  • Duplicated block (5 lines × 3) (crates/rig-core/src/providers/anthropic/options.rs)
  • Duplicated block (5 lines × 3) (crates/rig-core/src/providers/openai/options.rs)
  • Duplicated block (5 lines × 3) (crates/rig-tools/src/delete.rs)
  • Duplicated block (5 lines × 4) (crates/rig-core/src/providers/openai/options.rs)
  • …and 160 more

New (438)

  • AWS access key ID with no secret key beside it (crates/rig-core/src/redact.rs)
  • CacheBook::plan (cognitive 21) (crates/rig-core/src/providers/gemini/caching.rs)
  • CacheBook::plan (cyclomatic 20) (crates/rig-core/src/providers/gemini/caching.rs)
  • Duplicated block (10 lines × 2) (corpus/base/src/check_after_turn.rs)
  • Duplicated block (10 lines × 2) (corpus/base/src/custom_provider.rs)
  • Duplicated block (10 lines × 2) (corpus/ct/src/ask_session_cli/session.rs)
  • Duplicated block (10 lines × 2) (crates/rig-milvus/src/lib.rs)
  • Duplicated block (10 lines × 2) (test-support/rig-test-support/src/model_session.rs)
  • Duplicated block (10 lines × 3) (corpus/sn/src/terminal_sessions.rs)
  • Duplicated block (11 lines × 2) (crates/rig-ecs/src/turn.rs)
  • Duplicated block (12 lines × 2) (corpus/oc/src/dataset_cli/dataset.rs)
  • Duplicated block (13 lines × 2) (corpus/oc/src/dataset_cli.rs)
  • Duplicated block (13 lines × 2) (corpus/sn/src/terminal_sessions.rs)
  • Duplicated block (13 lines × 2) (plugins/rig-acp/src/router.rs)
  • Duplicated block (16 lines × 2) (corpus/ct/src/ask_session_cli.rs)
  • Duplicated block (2–5 lines × 3) (crates/rig-core/src/providers/openai/options.rs)
  • Duplicated block (5 lines × 2) (crates/rig-core/src/providers/gemini/extension.rs)
  • Duplicated block (5 lines × 2) (crates/rig-tools/src/list.rs)
  • Duplicated block (6 lines × 2) (corpus/base/src/auto_commit.rs)
  • Duplicated block (6 lines × 2) (corpus/base/src/pinned_notes.rs)
  • …and 418 more

Changes since last survey

  • 217 commits — 76 feature/other, 141 fixes

By area

  • crates/rig-core — 42 commits
  • crates/rig-cli — 27 commits
  • plugins/rig-basics — 18 commits
  • (root) — 16 commits
  • crates/rig-tools — 13 commits
  • crates/rig-harness — 12 commits
  • crates/rig-agent — 9 commits
  • crates/rig-ecs — 8 commits
  • crates/rig-memory — 6 commits
  • plugins/rig-acp — 6 commits
  • plugins/rig-mcp-client — 6 commits
  • plugins/rig-tui — 6 commits
  • crates/rig-cassette — 5 commits
  • crates/rig-harness-protocol — 5 commits
  • plugins/rig-claude-cloak — 4 commits
  • crates/rig-typesafeai — 3 commits
  • xtask/src — 3 commits
  • corpus/base — 2 commits
  • crates/rig-helixdb — 2 commits
  • crates/rig-lancedb — 2 commits

Notable commits

  • fix: fix!: every limit on asked work is a settings.toml key with a sensible default, and rig says what it changes (#2995)
  • fix: fix!: schema runs fail on answers that miss the schema, and gateway Claude ids take the model's facts (#3050)
  • fix: fix(agent)!: headless runs close what an exit cuts off, refuse unknown --tools, fail empty answers (#2915)
  • fix: fix(agent)!: headless runs hear SIGTERM in any setup, stop when nobody reads, and leave no tool processes behind (#2942)
  • fix: fix(agent)!: quitting mid-turn is a stop, and a restart carries nothing on by itself (#2943)
  • fix: fix(agent,xtask)!: answer batches are checked whole, dispatch patches keep the tool policy, the planner skips an empty service suite (#3097)
  • fix: fix(anthropic): nest a tool schema's root anyOf, oneOf or allOf under a double not (#3241)
  • fix: fix(anthropic): repoint references inside dependencies and contentSchema subschemas (#3246)
  • fix: fix(basics): a first pick names a saved key file it cannot read (#3168)
  • fix: fix(basics): a helper timeout too long for a deadline is refused (#3141)
  • fix: fix(basics): a saved /login key that can't be read refuses the connect (#3136)
  • fix: fix(basics): count a summary call a stop removes before it settles (#3167)
  • fix: fix(basics): settle a helper's count with try_update, which Rust 1.99 does not deprecate (#3215)
  • fix: fix(bedrock): place the message cache checkpoint after a reasoning turn on Claude models the catalog lists (#3103)
  • fix: fix(bedrock,copilot,openai,mistral): typed Bedrock calls, catalogue routing on Copilot, LiteLLM thinking_blocks, Codestral's width (#3047)
  • fix: fix(cassette): xAI recordings send store: false on Responses requests (#3090)
  • fix: fix(cassette,xtask,test-support): scrub team ids, price one-hour cache writes, close the guard's and registry's gaps (#3045)
  • fix: fix(catalog)!: refresh from models.dev, add GPT-6.1 Terra, Terra's Chat tool rule and the 12 h cache (#2956)
  • fix: fix(chatgpt): the browser sign-in opens gets no RIG_SECRET_ variable (#3171)
  • fix: fix(ci): main builds again after #3005 and #3014 crossed (#3040)
  • …and 197 more

Written by cca.canine.dev from the codebase's own history, inside the signed delivery this page is composed from.

Survey your own repository

0xPlaygrounds/rig was measured the same way every project in this corpus was: the same rubric, at a pinned commit, with the result published in full. Point a surveyor at a repository you know and see whether you agree with it.

About this page

  • The score is its most recent published measurement, taken on 14 November 2026 at a pinned commit. It is not a live figure and does not change until the project is measured again.
  • Measured at commit 8c1a2e19c8b253889e2d73c34e61898b6cb3c522 — the exact code this score is about.
  • Scored under rubric-2026.11.4 — the same rubric and the same method as every other entry in this index.
  • Measured by Canine Code Analyzer (cca.canine.dev) using cca measure-2026.11.65+preprod-e1a3b69fc8ee.