Skip to content
CAI
Software that uses CAICheck a score

DayuanJiang/next-ai-draw-io

55.5

Adequate · 7 November 2026

48.8k

lines of production code

TypeScript

primary language

4

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

Add multilingual support with locale-specific routing and metadata

The application now supports multiple languages (English, Simplified Chinese, Traditional Chinese, and Japanese) via dynamic routing under the app/\[lang\] directory. This change introduces locale-aware layouts that generate specific SEO metadata (titles, descriptions, Open Graph tags) and structured data for each language, while the main page component handles language persistence, automatic redirection based on saved preferences, and passes the current locale to the embedded Draw.io editor to ensure UI consistency.

app/\[lang\] · high confidence

Adds SEO metadata files and modernizes the application's visual theme

This change introduces dedicated metadata files for the application: \manifest.ts\ defines the PWA manifest (including icons and display settings), \robots.ts\ configures search engine crawling rules, and \sitemap.ts\ specifies the site's URL structure for indexing. Additionally, the global CSS (\globals.css\) has been updated to implement a new, modernized design system with refined color palettes for both light and dark modes, increased border radius, and the inclusion of the Tailwind Typography plugin, replacing the previous default styling.

app · high confidence

Canvas redesign with self-hosted draw.io and diagram versioning

The canvas area now uses a self-hosted draw.io editor (embedded via iframe) instead of the previous implementation, supporting same-origin direct control and cross-origin fallbacks with proper theme and locale handling. A new versioning system tracks diagram changes, providing a version strip with thumbnails, version cards with change summaries, and a compare dialog to view differences and restore previous states. Additionally, a 'My Templates' library has been added to the chat panel, allowing users to create, edit, pin, search, and manage custom prompt templates with import/export functionality. Also in this area: 2 commits (1 fix) modifying components.

components/canvas, components/chat · medium confidence

Electron desktop build pipeline and offline draw.io support

This change introduces a suite of build and development scripts to support a new Electron desktop application. The \prepare-electron-build.mjs\ and \afterPack.cjs\ scripts handle packaging the Next.js standalone output, ensuring symlinks are dereferenced to prevent macOS codesign failures, and performing ad-hoc signing. The \electron-dev.mjs\ script streamlines local development by orchestrating the Next.js and Electron processes. Additionally, \fetch-drawio.mjs\ and \drawio-zip.mjs\ enable bundling a specific version of the draw.io web app for offline use, while \update-model-catalog.mjs\ refreshes model pricing data from models.dev, and \indexnow.mjs\ automates search engine indexing notifications for the website.

scripts · high confidence

Expose server configuration via API endpoint

A new GET endpoint at /api/config has been added to the application, allowing clients to retrieve runtime configuration settings such as whether an access code is required, and the current daily request, token, and TPM limits directly from environment variables.

app/api/config · high confidence

File-based admin settings panel with secure provider management

A new file-based admin settings panel is available at /admin, allowing administrators to manage environment variables and AI provider configurations via a UI instead of editing .env files directly. The system uses a JSON settings file (defaulting to data/settings.json) that overlays onto process.env with higher precedence, supporting features like access codes, rate limits, and observability keys. Provider credentials are managed securely through the admin panel, with secrets masked in the UI and stored in ADMIN\_-prefixed environment variables to avoid shadowing standard keys. Authentication is enforced via the ADMIN\_PASSWORD environment variable, using timing-safe comparison for password verification.

lib/admin · high confidence

Initial Electron desktop application support

This change introduces the foundational structure for the Next AI Draw.io desktop application using Electron. It adds the \electron-builder\ configuration to define build targets (DMG, NSIS, AppImage, deb, rpm) and packaging rules, including bundling draw.io for offline support. It also implements the preload scripts and TypeScript type declarations that expose native capabilities to the web renderer, such as window controls, proxy settings management, user locale preferences, and configuration preset handling.

electron · high confidence

Initial i18n infrastructure and translations for English, Chinese, Japanese, and Traditional Chinese

The application now supports internationalization with a new localization layer in \lib/i18n\. This change introduces a configuration for supported locales (English, Simplified Chinese, Japanese, and Traditional Chinese) and a dictionary loading mechanism that fetches language-specific JSON files. It includes a utility for formatting localized messages with variables and provides the complete set of translation keys for the chat interface, settings, navigation, and error messages, enabling users to switch their interface language. Also in this area: Cleanup of unused internationalization strings and developer tools.

lib/i18n · high confidence

Initial project scaffolding and deployment configuration

The repository is initialized with a Next.js application structure, including a multi-stage Dockerfile for containerized builds, a docker-compose setup for local development with draw.io, and configuration files for Vercel, Cloudflare Workers, and EdgeOne Pages. The project adopts Biome for linting and formatting, integrates Langfuse for observability via an instrumentation hook, and supports subdirectory deployments through the NEXT\_PUBLIC\_BASE\_PATH environment variable. Additionally, an admin panel for runtime settings and i18n locale redirection middleware are included.

(repo-wide) · high confidence

Initial release of the Next AI Draw.io MCP server package

This change introduces the \packages/mcp-server\ directory, establishing the initial distribution package for the Next AI Draw.io MCP server. The package includes the server manifest (\server.json\ version 0.4.6), TypeScript configuration (\tsconfig.json\), and test configurations for Vitest and Playwright. It also bundles the necessary draw.io frontend assets (\drawio-files.txt\) to enable the live browser canvas preview and provides comprehensive installation and usage documentation (\README.md\) for AI agents like Claude Code, Cursor, and VS Code.

packages/mcp-server · high confidence

Introduce desktop application with configuration presets, proxy support, and i18n menus

This change introduces the Electron-based desktop application, establishing the core architecture for local configuration and UI. Users can now manage and switch between configuration presets (AI provider, model, API key, base URL, and temperature) which are securely encrypted using Electron's safeStorage API. The application supports proxy configuration via a settings window and persists user data (chats and settings) by binding to fixed local ports to maintain localStorage integrity across restarts. Additionally, the native application menu is fully internationalized, supporting English, Simplified Chinese, Japanese, and Traditional Chinese locales.

electron/main · high confidence

Introduce multi-provider model configuration types

Added \lib/types/model-config.ts\ to define the data structures for managing multiple AI providers and models. This includes the \ProviderName\ union type listing supported providers (such as OpenAI, Anthropic, Vertex AI, Ollama, and others), interfaces for \ModelConfig\, \ProviderConfig\, and \MultiModelConfig\ to handle per-provider settings and validation states, and metadata maps like \PROVIDER\_LOGO\_MAP\ and \PROVIDER\_INFO\ to support UI rendering and default configurations.

lib/types · high confidence

Introduces server-side model configuration and provider model listing endpoints

This change adds two new API routes to support dynamic model management. The new \app/api/server-models/route.ts\ endpoint allows the application to load and expose models configured via server environment variables (AI\_MODEL/AI\_PROVIDER), enabling Docker users to set these values at runtime. Additionally, \app/api/provider-models/route.ts\ provides a secure endpoint for clients to fetch available models from external providers; it requires an access code, validates the API key, and includes SSRF protection to block requests to private base URLs. These routes support the broader multi-provider and bring-your-own-API-key capabilities by providing the necessary infrastructure for model discovery and server-side configuration.

app/api/chat · high confidence

Introduction of new Radix-based UI components and refined button styling

This update adds a comprehensive set of new UI components to the application, including AlertDialog, Collapsible, Command (command palette), Dialog, Input, Label, Popover, Resizable panels, ScrollArea, Select, Switch, and Textarea, all built on top of Radix UI primitives. Additionally, the default button hover state has been changed from a lighter background to a darker brightness effect, providing a more distinct visual feedback when interacting with buttons.

components/ui · high confidence

Log diagram saves to Langfuse traces

The API now records when a user saves a diagram by creating a 'diagram-saved' score in Langfuse, linking the action to the most recent chat trace for the session. This allows users to track diagram save events alongside their conversation history in the observability platform.

app/api/log-save · high confidence

MCP server introduces compact cell format and custom drawing instructions

The MCP server now supports a compact XML format for shapes and edges (using x, y, w, h attributes) to reduce token usage in model outputs, with automatic expansion to standard draw.io format on the server side. It also adds support for user-defined drawing rules via a local \instructions.md\ file, which is appended to the system prompt to allow custom style and layout guidelines.

packages/mcp-server/src · high confidence

New EdgeOne Pages AI provider with OpenAI-compatible chat completions

Added a new edge function at \edge-functions/api/edgeai/chat/completions.ts\ that exposes an OpenAI-compatible API endpoint for EdgeOne Pages. This allows users to integrate with the AI SDK using \createOpenAI({ baseURL: '/api/edgeai' })\ and supports native tool calling via EdgeOne's \AI.chatCompletions\. The endpoint validates requests, enforces access codes via environment variables, and restricts model usage to a specific list of DeepSeek models.

edge-functions · high confidence

New MCP canvas shell for previewing diagrams

The MCP server now provides a dedicated canvas shell (served at /shell/) that displays the web app's draw.io canvas without the chat interface. This shell supports real-time synchronization with the server, allowing users to view version history cards, restore previous diagram states, and export diagrams. It includes a side panel for version navigation, handles dark/light theme preferences, and supports multiple languages (English, Chinese, Japanese).

packages/mcp-server/shell · high confidence

New MCP installation page with multi-client install tabs

A new localized page at /mcp provides installation instructions for the draw.io MCP server across multiple AI clients. It features a tabbed interface (InstallTabs component) that displays one-click install links for VS Code and Cursor, command-line instructions for Claude Code and Codex, and configuration snippets for other clients, ensuring all install steps are present in the HTML for better search engine visibility.

app/\[lang\]/mcp · high confidence

New URL content extraction API with SSRF protection and charset handling

The new /api/parse-url endpoint allows users to submit a URL and receive the extracted article title and content as Markdown. To ensure security, the endpoint blocks requests to private or internal IP addresses and strictly rejects HTTP redirects to prevent Server-Side Request Forgery (SSRF) attacks. It also includes robust handling for non-UTF-8 character encodings (such as Shift\_JIS or GBK) by detecting the charset from HTTP headers or HTML meta tags, and enforces limits on response size and content length to prevent resource exhaustion. PDF URLs are explicitly unsupported and return a clear error message.

app/api/parse-url · high confidence

New VLM-based diagram validation API endpoint

A new API endpoint at /api/validate-diagram has been added to support visual quality analysis of diagrams using a Vision Language Model (VLM). Users can submit a PNG image (as a Base64 data URL) to receive structured validation results, including identified issues and suggestions. The endpoint enforces a 5MB image size limit, integrates with the existing quota management system to track token usage against daily and per-minute limits, and includes cross-site request protection. Validation can be globally disabled via the ENABLE\_VLM\_VALIDATION environment variable, and the system gracefully handles missing model configurations or errors by returning a default valid result to avoid blocking user workflows.

app/api/validate-diagram · high confidence

New access code verification API endpoint

A new API route at /api/verify-access-code has been added to validate access codes server-side before any data is saved. This endpoint accepts POST requests and returns a 401 status with an error message if the code is invalid or missing, or a success response if valid, ensuring that access control is enforced at the API level.

app/api/verify-access-code · high confidence

New chat UI components for file handling, model configuration, and exports

This change introduces a suite of new React components that power the updated chat interface and settings experience. Users can now upload and preview images, PDFs, and text files via the new FilePreviewList component, which handles extraction status and character counts. Model management is handled by the ModelConfigDialog and ModelSelector, allowing users to configure provider credentials, validate API keys, and select from grouped server or user-defined models. The interface also includes a SaveDialog for exporting diagrams in multiple formats (drawio, PNG, SVG, xmlsvg) with options for transparent backgrounds, and a QuotaLimitToast to notify users of daily request or token limits. Supporting these features are new utilities like CodeBlock for syntax highlighting, a custom Image component for base-path handling, and a DevXmlSimulator for testing XML streaming.

components · high confidence

New example diagrams and static asset optimizations

The public folder now includes new example SVG diagrams, specifically an Agile Scrum workflow and a Transformer Architecture visualization, which serve as visual references for users. Additionally, a \\_headers\ file has been added to configure long-term immutable caching for Next.js static assets, improving load performance for these resources.

public · high confidence

New file-based admin settings panel at /admin

A new admin interface is now available at /admin, allowing administrators to manage model configurations and general application settings via a file-based backend. The panel includes a models section for configuring provider credentials, selecting default models, and testing connectivity, as well as a general settings section that displays the source of each configuration (file, environment, or default) and supports toggling feature groups. The interface is secured with a password, persists the session in sessionStorage, and includes a warning when leaving the page with unsaved changes. Additionally, the admin layout explicitly sets metadata to prevent search engine indexing (noindex, nofollow).

app/\[lang\]/admin · high confidence

New file-based admin settings panel with provider management and connection testing

This change introduces a new admin interface at /admin that allows administrators to manage AI provider configurations and general settings via a file-based backend. The new API endpoints enable reading and updating provider definitions (including handling environment variable fallbacks and secret masking), managing a registry of general settings with type validation, and testing model connections using the current provider state. The implementation includes security measures to prevent secret exfiltration by validating endpoint consistency when resolving masked secrets, and ensures that settings changes are persisted to a writable file while respecting environment variable overrides.

app/api/admin · high confidence

New hooks for diagram tools, model configuration, and session management

The application introduces several new React hooks to support recent feature additions. useDiagramToolHandlers manages the logic for diagram-related AI tools (display, edit, append), including VLM-based validation and canvas state tracking. useModelConfig provides a client-side interface for multi-provider model settings, persisting configurations to localStorage and syncing with server-side model lists. useSessionManager handles chat session persistence via IndexedDB, including save/load operations, session switching, and concurrency guards. Additional hooks include useValidateDiagram for standalone diagram validation, useDictionary for i18n context, usePanelWidth for responsive layout adjustments, usePathLocale for URL-based locale detection, and useSessionManager for session lifecycle management.

hooks · high confidence

The site now includes a suite of new components to enhance the user experience: a \DemoPlayer\ that scripts an interactive tour of the editor's features (generation, hand-editing, AI editing, and version history) with localized text; a \DocMarkdown\ renderer that integrates these demos and \CopyCommand\ buttons directly into documentation pages; a \Gallery\ component displaying 23 AI-generated diagrams with prompts and category filters; and a \DocsToc\ for navigation. These additions support the new static site pages, including the landing page, docs, and MCP server information, with full support for English, Simplified Chinese, Japanese, and Traditional Chinese.

components/site · high confidence

New model validation API endpoint with SSRF protection

A new API route at /api/validate-model has been introduced to allow users and the admin panel to verify that a specific AI provider, model, and credentials are working correctly before use. This endpoint supports a wide range of providers (including AWS Bedrock, Vertex AI, Ollama, and EdgeOne) and enforces strict security measures, specifically blocking Server-Side Request Forgery (SSRF) attacks by preventing requests to private or internal URLs unless explicitly authorized by admin authentication. It also integrates with the existing quota management system to count validation attempts against daily request limits for non-admin users.

app/api/validate-model · high confidence

New multilingual landing page for Next AI Draw.io

A new localized landing page is now available at /about/\[lang\], featuring a redesigned layout with a hero section, a demo player, and three usage pathways (browser, desktop, and AI agent). The page includes a gallery of example diagrams, detailed feature descriptions, and provider information, with all text content translated for English, Chinese, and Japanese locales.

app/\[lang\]/about · high confidence

New provider model listing endpoint with security validations

A new API route at app/api/provider-models has been added to handle fetching available models for a given provider. This endpoint validates that the request body contains only string values for the API key and base URL, rejecting non-text credentials to prevent type-based injection issues. It also enforces SSRF protection by blocking private base URLs unless explicitly allowed, and requires an access code for untrusted client-specified URLs. The endpoint returns a list of models or a null value for providers that do not support model listing, ensuring that internal error details are not exposed to the client.

app/api/provider-models · high confidence

New security, configuration, and infrastructure libraries

This change introduces several new server-side libraries that enhance security, deployment flexibility, and observability. The \lib/access-code.ts\ module adds cross-site request protection (requiring JSON content types and validating Host headers in desktop mode) and optional access-code authentication via the \x-access-code\ header. \lib/base-path.ts\ enables subdirectory deployments by dynamically prefixing API and asset URLs based on the \NEXT\_PUBLIC\_BASE\_PATH\ environment variable. \lib/dynamo-quota-manager.ts\ implements server-side daily quota tracking using a DynamoDB composite key schema (user ID and date) with configurable timezones and limits for requests, tokens, and TPM. Additionally, \lib/langfuse.ts\ restores and simplifies Langfuse observability integration for chat traces, while \lib/llm-errors.ts\ provides structured error classification for LLM providers to improve user-facing error messages.

lib · high confidence

New workspace UI components and layout

The workspace area now includes a new set of UI components: a toast notification container (AppToaster), a brand mark icon (BrandMark), a documentation link button (DocsButton), and a reusable icon button with tooltips (IconButton). These are integrated into the main Workspace component, which now features a StartBar with the app name, docs, and settings buttons, and handles mobile view logic for the canvas and chat panel.

components/workspace · high confidence

User feedback is now linked to chat traces in Langfuse

The new /api/log-feedback endpoint allows user feedback (good/bad) to be recorded in Langfuse. When a session ID is provided, the system attempts to attach the feedback score to the most recent chat trace for that session; if no trace exists, it creates a standalone feedback trace. Input is validated using Zod, and the endpoint gracefully handles cases where the Langfuse client is unavailable or no session ID is provided.

app/api/log-feedback · high confidence

Removals

Removed assistant-ui component library files

Deleted the \components/assistant-ui\ directory, removing the custom React components for the chat interface including \markdown-text.tsx\ (for rendering markdown with code copy support), \thread-list.tsx\ (for managing conversation threads), \thread.tsx\ (for the main chat viewport and message composition), and \tooltip-icon-button.tsx\ (a shared UI utility).

components/assistant-ui · high confidence

Behavioural changes

2 commits (1 fix) modifying lib/drawio

A change to existing behaviour in lib/drawio — 2 commits (1 fix), 1 file: “feat(canvas): a style button next to the selection opens draw.io's format panel (\#1033) (\#1039)”; “fix(desktop): save draw.io PDF exports with printToPDF (\#1035) (\#1038)” (refs \#1033, \#1039, \#1035, \#1038). Also in this area: Improved diagram version tracking and AI response behavior; Removal of configuration presets and settings window; Canvas supports transient previews and AI selection details; AI diagram edits now target the specific page the user is viewing.

lib/drawio · medium confidence · unverified

3 commits (1 fix) modifying app/\[lang\]/docs

A change to existing behaviour in app/\[lang\]/docs — 3 commits (1 fix), 1 file: “feat: a docs button on the start screen, the docs text in the site's colors, the same header on every site page (\#1010)”; “fix(site): index the English pages under their own URLs, 404 instead of 500 for /llms.txt, AWS page wording (\#1020)”; “feat(site): landing page at /about, an MCP server page and a docs site in four languages (\#1008)” (refs \#1010, \#1020, \#1008).

app/\[lang\]/docs · medium confidence · unverified

Added build scripts to ensure npm package includes required assets

The MCP server package now includes two new scripts, \copy-assets.mjs\ and \check-package.mjs\, to guarantee that non-TypeScript assets (such as shape library documentation and the browser preview page) are correctly copied to the distribution directory and included in the published npm package. The \copy-assets.mjs\ script handles the copying of these assets after the TypeScript compilation, while \check-package.mjs\ verifies that all required files are present in the package output, specifically handling differences in \npm pack\ output formats between npm 11 and 12.

packages/mcp-server/scripts · high confidence

Centralized diagram state management with validation and export controls

The diagram context has been refactored into a single, centralized provider that manages chart XML, SVG state, and history. This change introduces pre-display XML validation to prevent loading malformed diagrams and adds specific export capabilities, including XML/SVG and PNG capture for validation purposes. It also standardizes the handling of diagram history and save dialogs, ensuring consistent state across the application.

contexts · high confidence

Enforce linting, type-checking, and tests via Husky hooks

The project now automatically runs lint-staged and TypeScript type-checking before every commit, and executes the test suite before every push (skipped if node\_modules are missing). This ensures code quality and test coverage are maintained without manual intervention.

.husky · high confidence

New Zustand-based state management for canvas, settings, UI, and versions

The application now uses dedicated Zustand stores to manage core state, replacing previous implementations. The new \canvas-store\ tracks the draw.io editor state, selection, and busy indicators. \settings-store\ handles user preferences (theme, shortcuts, panel width) with local storage persistence and a pre-render script to prevent dark-mode flash. \ui-store\ manages UI layout state, including the chat panel, mobile views, and settings navigation. \versions-store\ tracks diagram version history, supporting undo/redo, version capping, and session persistence.

stores · high confidence

Test coverage

Added end-to-end tests for chat, diagrams, and error handling; Added tests for MCP server core logic and integration; Added unit test snapshots for provider configuration matrix.

Dependencies

Major dependency upgrades and new MCP server package

The application dependencies have been significantly upgraded, including Next.js to v16, React to v19.1.2, the AI SDK to v6, and Zod to v4, alongside the addition of numerous new AI provider integrations (Amazon Bedrock, Azure, DeepSeek, Google Vertex, OpenRouter, etc.) and observability tools (Langfuse, OpenTelemetry). A new \@next-ai-drawio/mcp-server\ package has been introduced at version 0.3.0, providing a standalone Model Context Protocol server with its own dependencies (including Zod v4 and Open v11) and build scripts, distributed via \next-ai-drawio-mcp\. The main app also now includes Electron support for desktop distribution and Cloudflare Workers deployment options.

(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

This is the PUBLIC form of this artifact. Findings are listed in full, but the details of SECURITY findings — which rule fired, in which file, on which line, and how to fix it — are deliberately withheld, and any secret-scanner results are excluded entirely. Where detail is absent here it was REMOVED FOR PUBLICATION; it is not missing from the analysis. The complete artifact is available from the repository owner.

Score

  • CAI 51 → 55 (+4.2)
  • Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

Lenses

  • Code Health 61 → 60 (-0.7)
  • Architecture 91 → 90 (-1.0)
  • Maturity 51 → 77 (+26.0)
  • Readiness 48 → 51 (+3.1)
  • Security 75 → 79 (+3.8)
  • Accessibility 49 → 53 (+3.4)
  • Performance 60 → 60 (+0.0)

Resolved (58)

  • High CVE: [GHSA redacted] (packages/mcp-server/package-lock.json)
  • High CVE: [GHSA redacted] (package-lock.json)
  • High CVE: [GHSA redacted] (package-lock.json)
  • High CVE: [GHSA redacted] (packages/mcp-server/package-lock.json)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • Medium CVE: [GHSA redacted] (package-lock.json)
  • Off-boarding risk: anonymized user #1
  • Outdated (npm): @ai-sdk/amazon-bedrock
  • Outdated (npm): @ai-sdk/anthropic
  • Outdated (npm): @ai-sdk/azure
  • Outdated (npm): @ai-sdk/deepseek
  • Outdated (npm): @ai-sdk/google
  • Outdated (npm): @ai-sdk/google-vertex
  • Outdated (npm): @ai-sdk/openai
  • Outdated (npm): @ai-sdk/openai-compatible
  • Outdated (npm): @ai-sdk/react
  • Outdated (npm): @aws-sdk/client-dynamodb
  • …and 38 more

New (51)

  • Dependency hygiene PARTLY measured — npm pinning read, dependency currency not (registry.npmjs.org unreachable)
  • Deprecated: @opentelemetry/sdk-trace-base (package-lock.json)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • High: security finding (details withheld)
  • Hotspot: components/canvas/selection-ask.tsx (components/canvas/selection-ask.tsx)
  • Hotspot: packages/mcp-server/shell/canvas-shell.tsx (packages/mcp-server/shell/canvas-shell.tsx)
  • Medium CVE: [GHSA redacted] (package-lock.json)
  • Medium CVE: [GHSA redacted] (package-lock.json)
  • …and 31 more

Changes since last survey

  • 31 commits — 18 feature/other, 13 fixes

By area

  • (root) — 11 commits
  • packages/mcp-server — 8 commits
  • .github/workflows — 3 commits
  • lib/i18n — 2 commits
  • (repo) — 1 commit
  • docs/cn — 1 commit
  • electron/main — 1 commit
  • lib/model-catalog.json — 1 commit
  • packages/dsh-plugin — 1 commit
  • scripts/indexnow.mjs — 1 commit
  • tests/unit — 1 commit

Notable commits

  • fix: fix(chat): a saved prompt fills the message box instead of sending it (#1030) (#1036)
  • fix: fix(chat): five failures from the demo's traces: 10 MB requests, OpenAI's official URL, image errors, Claude thinking, empty files (#1066)
  • fix: fix(deps): bump the npm_and_yarn group across 2 directories with 9 updates (#1063)
  • fix: fix(deps): update minor and patch dependencies (#1056)
  • fix: fix(desktop): save draw.io PDF exports with printToPDF (#1035) (#1038)
  • fix: fix(indexnow): sort the head tags and leave out Next's font tag before comparing (#1044)
  • fix: fix(mcp-server): count coordinates of 0 as left out, so one undo takes back the AI's last edit, release 0.4.10 (#1049)
  • fix: fix(mcp-server): keep a shape on layer 1 when another layer comes first, release 0.4.7 (#1037)
  • fix: fix(mcp-server): open the preview when a chat glues text onto its link, release 0.4.9 (#1048)
  • fix: fix(mcp-server): read a page's cells in draw.io's order, so a drawing with containers is not taken for a hand edit, release 0.4.11 (#1053)
  • fix: fix(mcp-server): repeat the preview link when the preview tab is not open, release 0.4.12 (#1062)
  • fix: fix(seo): permanent locale redirect for pages, nofollow on the gallery's prompt links (#1031)
  • fix: fix: keep the MiniMax M3.1 Flash change to the catalog and the suggestion list
  • change: Merge pull request #1026 from octo-patch/octopatch/parameter-refresh-minimax-m3
  • change: chore(deps): update dependency @biomejs/biome to v2.6.6 (#1051)
  • change: chore(deps): update dependency @modelcontextprotocol/sdk to v1.32.0 [security] (#1028)
  • change: chore(deps): update dependency @opennextjs/cloudflare to v1.20.2 [security] (#1000)
  • change: chore(deps): update dependency electron to v42 [security] (#1029)
  • change: chore(deps): update dependency electron to v45 (#1057)
  • change: chore(deps): update dependency electron-builder to v26.18.0 (#1052)
  • …and 11 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

DayuanJiang/next-ai-draw-io 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 7 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 04b1de949f9f2f14971bf2e26f4a7fd8cc392626 — the exact code this score is about.
  • Scored under rubric-2026.11.2 — 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-2026.11.3+preprod-110e4fee044f.