agentjido/jido_ai

Jido.AI is an Elixir-based framework for building and managing AI agents, providing a structured system for LLM interactions, planning, and reasoning. It supports multiple reasoning strategies such as Chain of Thought, Tree of Thoughts, and Graph of Thoughts, and allows for dynamic tool registration and skill-based extensibility. The system includes a CLI for execution, comprehensive testing infrastructure, and robust error handling and telemetry for agent lifecycles.

This change introduces the foundational structure for the Jido.AI v2.0.0-rc.0 release, including the \AGENTS.md\ guide, \CHANGELOG.md\, and \CONTRIBUTING.md\. It adds configuration files for quality gates (\.doctor.exs\, \coveralls.json\, \dialyzer.ignore-warnings\), an example environment file (\.env.example\), and updates the \README.md\ to reflect the new v2 architecture, installation via Igniter, and usage patterns. The \LICENSE\ file is also added/updated to Apache 2.0 with the correct copyright holder.

(repo-wide) · high confidence

The \lib/jido\_ai\ library now includes a comprehensive set of new action modules for direct LLM interactions, goal planning, and text analysis. Users can now execute chat completions (\llm\_chat\), text generation (\llm\_complete\), structured JSON object generation (\llm\_generate\_object\), and text embeddings (\llm\_embed\) with built-in model alias resolution, input validation, and telemetry. Additionally, new planning actions (\planning\_decompose\, \planning\_plan\, \planning\_prioritize\) allow agents to break down complex goals, generate structured plans with constraints, and prioritize tasks. Reasoning capabilities are expanded with \reasoning\_analyze\ for sentiment, topic, and entity extraction. These actions share a common helper infrastructure for input sanitization, option building, and usage tracking, providing a standardized interface for integrating LLM capabilities into agent workflows.

_lib/jido\ai · high confidence

This change introduces four new Mix tasks to the \lib/mix\ directory. \mix jido\_ai\ provides a unified command-line interface for running AI queries in one-shot or batch (stdin) modes, supporting various agent types, models, and output formats. \mix jido\_ai.install\ automates the setup of Jido AI within an application using Igniter, configuring default model aliases and prompting for API keys. \mix jido\_ai.skill\ allows users to list, show, and validate AI skills from directories or specific files. Finally, \mix jido\_ai.quality\ offers a maintainer-focused checkpoint that runs fast and full test gates and verifies traceability closure against a story matrix.

lib/mix · high confidence

The \examples/\ folder now includes several new demonstration agents and supporting tools. The \ApiSmokeTestAgent\ allows iterative HTTP debugging using GET and POST tools. The \BrowserAgent\ enables web browsing, searching via Brave, and page snapshotting. The \IssueTriageAgent\ demonstrates secure GitHub issue management with environment-based token injection. Additional examples include \TaskListAgent\ for iterative planning, \ReleaseNotesAgent\ using Graph-of-Thoughts synthesis, \WeatherAgent\ for location-based forecasts, \CalculatorAgent\ for arithmetic, and \SkillsDemoAgent\ showcasing module and file-based skills. Compatibility fallbacks for legacy weather tools are also provided.

examples · high confidence

The application now uses a new configuration structure (config/config.exs, dev.exs, test.exs) that standardizes logger metadata to include Jido.AI context and error details. In development, git hooks are configured to enforce commit message formatting via git\_ops and check code formatting on push, while test environments suppress most logs and disable auto-installed hooks.

config · high confidence

The core library module has been renamed from JidoAi to Jido.AI, bringing with it a comprehensive set of new capabilities for managing AI interactions. Users can now configure and resolve semantic model aliases (such as :fast or :capable) instead of hardcoding provider strings, and set role-based LLM defaults for text, object, and streaming generation. The module also exposes runtime tool management APIs to register, unregister, and list tools dynamically for running agents. As part of this restructuring, the example Politician action has been removed from the library.

lib · high confidence

Added comprehensive test coverage for Jido.AI core components; Added concurrency and migration tests for AI quota and retrieval stores; Added integration tests for plugin lifecycle, schemas, and signal routing; Added test coverage for CLI reasoning agent adapters and mix task contract; Added test coverage for ReAct reasoning runtime components; Added test coverage for the Jido.AI.Skill system; Added test coverage for the TRM reasoning engine components; Added test support utilities for CLI, directives, and LLM mocking; Added tests for AI signal helper functions; Added tests for Chain of Thought reasoning plugin; Added tests for Graph-of-Thoughts machine logic and structure enforcement; Added tests for ReAct action helpers and runtime actions; Added tests for Tree-of-Thoughts reasoning machine and structure enforcement; Added tests for directive deadlock prevention and error handling; Added tests for error model normalization and message sanitization; Added tests for example agents and weather tools; Added tests for reasoning strategy entrypoints, helper observability, and request lifecycle; Added tests for reasoning strategy plugins; Added tests for skill loading and resource access actions; Added tests for the Planning plugin's specification, initialization, and action routing; Added tests for the Reasoning plugin and its security validations; Added unit and integration tests for AI reasoning strategies; Added unit tests for LLM skill actions and plugin configuration; Added unit tests for Planning skill actions; Added unit tests for reasoning skill actions and strategy execution; Added unit tests for tool-calling actions; Expanded integration test coverage for Jido.AI v2 migration and strategy surface; Expanded test coverage for Jido.AI module and test infrastructure updates.

This release bumps the project version to 2.3.0 and aligns the baseline to Elixir 1.18. It upgrades the core Jido ecosystem dependencies (jido, jido\_action, req\_llm) to their 2.3/1.14+ ranges and introduces several new runtime dependencies including fsmx, splode, yaml\_elixir, and zoi. Development tooling is updated with newer versions of credo, dialyxir, ex\_doc, and mimic, while the project now includes comprehensive test coverage configuration via ExCoveralls and dialyzer PLT settings.

(dependencies) · high confidence

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

• First survey — no prior run to compare against. CAI 71.

• Code Health 93
• Architecture 99
• Maturity 64
• Readiness 72
• Security 74

• 300 commits — 210 feature/other, 90 fixes

• lib/jido_ai — 139 commits
• (root) — 74 commits
• (repo) — 30 commits
• test/jido_ai — 25 commits
• .github/workflows — 7 commits
• guides/developer — 5 commits
• guides/user — 5 commits
• lib/examples — 3 commits
• lib/jido_ai.ex — 2 commits
• lib/mix — 2 commits
• .agents/skills — 1 commit
• .github/dependabot.yml — 1 commit
• examples/lib — 1 commit
• examples/scripts — 1 commit
• notes/features — 1 commit
• specs/stories — 1 commit
• test/test_helper.exs — 1 commit
• todo/core — 1 commit

• fix: Fix missing log_level propagation in Turn.execute_internal/6 (#241)
• fix: Merge branch 'develop' into feature/fix-dialyzer-warnings
• fix: Merge branch 'feature/fix-dialyzer-warnings'
• fix: Merge branch 'main' into codex/fix-146-registry-lifecycle-consistency-merge
• fix: Merge pull request #149 from agentjido/codex/fix-138-execute-tool
• fix: Merge pull request #150 from agentjido/codex/fix-139-tool-calling-mount-schema
• fix: Merge pull request #151 from agentjido/codex/fix-140-skill-registry-not-started
• fix: Merge pull request #152 from agentjido/codex/fix-141-docs-api-config-alignment
• fix: Merge pull request #153 from agentjido/codex/fix-142-start-stream-supervisor-context
• fix: Merge pull request #155 from agentjido/codex/fix-143-callback-wrapper-supervisor-safety
• fix: Merge pull request #156 from agentjido/codex/fix-144-request-await-doc-options
• fix: Merge pull request #157 from agentjido/codex/fix-145-list-tools-zoi-schema-serialization
• fix: Merge pull request #158 from agentjido/codex/fix-146-registry-lifecycle-consistency
• fix: Merge pull request #159 from agentjido/codex/fix-147-quality-gate-strictness
• fix: fix(adaptive): preserve compatibility strings for failures (#234)
• fix: fix(agent): accept full ReqLLM model inputs (#206)
• fix: fix(agent): evaluate module attribute AST for system_prompt option (#217)
• fix: fix(agent): normalize CoT and CoD system_prompt handling (#218)
• fix: fix(audit): resolve actionable deep-audit findings
• fix: fix(context): preserve file reference media types (#306)
• …and 280 more

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

agentjido/jido_ai 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.