Skip to content
CAI
Software that uses CAICheck a score

sickn33/agentic-awesome-skills

40.8

Weak · 15 November 2026

392.5k

lines of production code

Python

with JavaScript

5

measurements over time

CAI band scale
CAI trend line
CAI lens gauges

What this system is

Features

Add 007 security audit and hardening skill

Introduces the '007' skill, a comprehensive security audit and hardening tool for AI agents. It provides structured threat modeling using STRIDE and PASTA frameworks, automated security scanning via Python scripts, and detailed incident response playbooks for data breaches, DDoS, ransomware, and supply chain compromises. The skill includes reference guides for API security, prompt injection defenses, and agent isolation patterns to help users identify and mitigate critical vulnerabilities in their projects.

skills · high confidence

Add editorial bundle documentation template

A new template file for editorial skill bundles has been added to the tools/templates directory. This template generates documentation for curated collections of skills organized by role and expertise level, providing users with quick-start instructions for installation across various AI assistants (Claude Code, Codex CLI, Cursor, Gemini CLI), guidance on selecting and combining bundles for specific goals (such as shipping an MVP or hardening production apps), and structured learning paths for different technical domains.

tools/templates · high confidence

Added Chinese-language documentation and licensing information

This change introduces the initial Chinese localization for the project's documentation. It adds a Chinese translation of the Microsoft MIT license (noting it is for reference only) and a comprehensive 'Sources and Attribution' page (\sources.md\) that catalogs official and community skills, their original sources, and licensing details. This provides Chinese-speaking users with access to project legal terms and skill provenance information in their language.

_docs\zh-CN/sources · high confidence

Backend implementation for the todo example app

The backend for the todo-app-generated example now includes a complete Express.js server with SQLite persistence. This adds a REST API for managing todos (create, read, update, delete) with input validation and rate limiting, alongside database initialization and migration logic using better-sqlite3.

skills/loki-mode/examples/todo-app-generated/backend · high confidence

Explainable auto-categorization for skills

The skill index now automatically classifies uncategorized skills into meaningful categories (such as Backend, Web Development, DevOps, and AI/ML) using keyword analysis and dynamic ID-based inference. This change introduces two new fields, \category\_confidence\ and \category\_reason\, to every record in \skills\_index.json\, allowing users to understand how and why a category was assigned. The system prioritizes explicit frontmatter categories, then falls back to keyword matching against a defined set of terms, and finally derives a category from the skill ID if no keywords match. This results in a more organized catalog where skills are grouped by relevance rather than left as 'uncategorized'.

_skill\categorization · high confidence

Initial release of the AAS Agent & MCP Builder plugin bundle

The 'AAS Agent & MCP Builder' bundle is now available as a plugin for Claude Code, Codex, and the agent-plugins ecosystem (version 18.13.0). This bundle provides a set of skills for building and evaluating agentic applications, including agent architecture design, LangGraph workflow construction, MCP tool building, context window management, and Langfuse observability. It also includes an agent evaluation skill with versioned case handling and uncertainty reporting, along with detailed architectural sketches for RAG pipelines and agent patterns.

plugins/agentic-bundle-aas-agent-mcp-builder · high confidence

Introduce AAS v1 MCP server with strict JSON parsing and bounded request handling

This change adds the AAS v1 Model Context Protocol (MCP) server implementation, providing a new stdio-based interface for agent interactions. It introduces a custom strict JSON parser that enforces a 16-level depth limit, rejects duplicate keys, and prevents JSON-RPC batch requests to mitigate parsing-related risks. The server exposes tools for skill search, stack composition, and selection evidence inspection, while enforcing strict byte limits (4 KB base, 256 KB line) and a 32-request concurrency cap to ensure stable, bounded performance.

tools/lib/aas-v1/mcp · high confidence

Introduce AAS v1 adapters for Claude and Codex configuration management

Added a new adapter layer in \tools/lib/aas-v1/adapters\ that manages configuration for the AAS agent-first control plane on Claude and Codex hosts. The \claude.js\ adapter parses and writes JSON-based MCP server settings, while \codex.js\ handles TOML-based configuration, including support for inline comments and array syntax. Both adapters enforce strict input validation, detect unknown configuration keys, and ensure file integrity via SHA-256 digests. The \safety.js\ module adds platform-specific security controls, including Windows ACL hardening, ownership verification, and safe directory checks, while \values.js\ normalizes server command, arguments, and environment variables before they are written to disk.

tools/lib/aas-v1/adapters · high confidence

Introduce AAS v1 library with secure catalog, search, and stack composition

The \tools/lib/aas-v1\ directory now contains a new library that provides a secure, auditable foundation for skill catalog management and stack composition. It introduces canonical JSON serialization and SHA-256 hashing to ensure deterministic data handling, alongside a robust catalog loader that verifies offline asset integrity against a signed manifest. The library includes a search engine with strict input validation and neutral filtering, a durability module that ensures filesystem consistency on Windows via explicit directory flushing, and a selection engine that composes stacks with explicit, auditable evidence of the selection process. This change adds the core infrastructure for safe, verifiable skill discovery and stack building.

tools/lib/aas-v1 · high confidence

Introduce Antigravity Web App as a static skill catalog and local workbench

The repository now includes a new React-based web application (\apps/web-app\) that serves as a static catalog and browser for the \agentic-awesome-skills\ library. This app renders searchable skill detail pages, home, category, and bundle routes, and publishes the public site to GitHub Pages. It supports a local-only "refresh skills" developer flow via a Vite dev server plugin, which is hidden in production unless explicitly enabled. The application treats save/star interactions as browser-local by default, with optional read-only Supabase integration for community save counts. It also includes a local workbench mode for plugin development, configured via a separate build target.

apps/web-app · high confidence

Introduce fully local AAS plugin with offline Workbench review

Users can now install the Agentic Awesome Skills (AAS) plugin as a self-contained, local stdio package that runs entirely on their machine without requiring an HTTP server, API keys, or network access. The plugin bundles the complete AAS catalog, a read-only Core MCP runtime, and a Workbench UI for reviewing agent-composed stacks and evidence. It exposes three native workflow skills—discovery, composition, and review—along with an \open\_workbench\ tool that allows users to inspect manifests and selection evidence locally. The Workbench component in the web app has been updated to support this embedded mode, validating host artifacts in browser memory without fetching external resources.

apps/aas-plugin, apps/web-app/src/plugin · high confidence

Introduce new CLI, installer, and MCP entry points

Added three new executable scripts in tools/bin: aas.js (the main CLI entry point), install.js (the skill installer supporting multiple targets like Cursor, Claude, and Antigravity), and aas-mcp.js (the MCP server entry point with verified catalog resolution). These files establish the runtime interfaces for the Agentic Awesome Skills tooling.

tools/bin · high confidence

Introduce safe, atomic git push workflow with isolated staging

A new 'git-pushing' skill and its accompanying \smart\_commit.sh\ helper are added to safely stage, commit, and push changes. The tool enforces conventional commit messages, prevents mixed-index commits by staging in an isolated temporary index, and uses file locking to ensure atomic branch updates, rejecting operations on detached HEADs or non-'files' ref backends.

plugins/agentic-bundle-essentials/skills/git-pushing · high confidence

Introduce skill-creator for automated CLI skill generation

The skill-creator is now available in the \plugins/agentic-bundle-skill-author/skills/skill-creator\ directory, providing an automated workflow for generating new CLI skills for GitHub Copilot CLI and Claude Code. This new capability includes a Python-based initializer (\scripts/init\_skill.py\) to scaffold skill structures, a packager (\scripts/package\_skill.py\) to create distributable \.skill\ zip files, and a validation script (\scripts/quick\_validate.py\) to enforce metadata and content standards. The skill guides users through interactive brainstorming, template application, and installation, supported by detailed reference documentation and output patterns.

plugins/agentic-bundle-skill-author/skills/skill-creator · high confidence

Local skill shortlisting and saving with cross-tab sync

The web app now supports browser-local management of skill shortlists and saved skills (stars) via the new \useSkillShortlist\ and \useSkillStars\ hooks. Users can toggle skills in their shortlist and mark skills as saved, with changes persisted to \localStorage\ and automatically synced across open browser tabs and multiple hook instances. The implementation includes robust error handling for storage failures, corrupted data, and legacy storage keys, ensuring the UI remains stable even if persistence is unavailable. Additionally, a new \usePageMeta\ hook allows components to declaratively update SEO metadata, and a read-only Supabase client is exposed for community save count feeds.

apps/web-app/src/hooks · high confidence

New 007 security audit skill with comprehensive reference materials

Added a new '007' skill for security auditing, hardening, and threat modeling (STRIDE/PASTA), including detailed reference guides for AI agent security, API security patterns, OWASP checklists (Web, API, LLM), incident response playbooks, and a Python-based configuration hub for scoring and verdicts. Also added an 'Andru.ia Solutions Architect' skill for project diagnosis and roadmap planning in Spanish.

plugins/agentic-awesome-skills-claude/skills · high confidence

New AAS Preview verification tooling for agent-first assurance

The verification/aas-preview directory now includes a suite of Node.js scripts (aggregate, installation, run-installed-candidate, runner, write-workbench-receipt) that implement the verification pipeline for the agent-first preview assurance profile. These tools validate installation receipts, exercise the CLI and copier on Windows (PowerShell 5.1 and 7) and POSIX shells, enforce write-guard policies, verify MCP local-stdio and read-only snapshot contracts, and aggregate cross-platform job results into a single bundle, ensuring that preview builds meet the agent-first-preview-1 assurance requirements before release.

verification/aas-preview · high confidence

New AI product evaluation and analytics skills added

Added a suite of new skills to the \skills\ directory to support AI product development and evaluation workflows. The \ab-test-setup\ skill provides a structured procedure for designing A/B tests, including hypothesis locking, sample size estimation, and tracking verification. The \agent-evaluation\ skill introduces versioned case-based evaluation with explicit verifiers and uncertainty reporting for comparing agent or prompt changes. The \hugging-face-evaluation\ skill enables adding structured evaluation results to Hugging Face model cards, supporting extraction from READMEs, imports from Artificial Analysis, and custom evaluations using vLLM or lighteval. Additionally, new skills for \ai-wrapper-product\ (building focused AI tools), \analytics-product\ (PostHog/Mixpanel analytics and retention analysis), \context-window-management\ (LLM context optimization strategies), and \kpi-dashboard-design\ (KPI framework and dashboard patterns) have been introduced to guide users through specific product and operational tasks.

plugins/agentic-bundle-aas-ai-product-evaluation-ops/skills · high confidence

New API design and testing skills added to the Agentic Awesome Skills bundle

The \plugins/agentic-bundle-aas-api-platform-builder/skills\ directory now includes a comprehensive set of new documentation-based skills for API development. These include \api-contract-testing\ for verifying provider/consumer compatibility, \api-design-principles\ with supporting checklists and reference guides for REST and GraphQL, \api-documentation\ for OpenAPI and developer guide workflows, \api-endpoint-builder\ for generating production-ready endpoint code, and \api-patterns\ which provides decision trees for API style selection (REST, GraphQL, tRPC) along with a heuristic triage script (\api\_validator.py\) for local code scanning. These additions provide structured guidance and tooling for designing, documenting, and validating APIs within the platform.

plugins/agentic-bundle-aas-api-platform-builder/skills · high confidence

New Agentic Skills for Analytics, Market Research, and Content Creation

This update adds several new community-sourced skills to the Agentic Awesome Skills bundle. The analytics-tracking skill provides a framework for designing and auditing measurement systems, including a diagnostic readiness index. The apify-market-research skill enables automated market analysis by integrating with Apify Actors to extract data from platforms like Google Maps, Facebook, and TripAdvisor, supported by a new Node.js runner script. The content-creator skill expands content capabilities with local Python scripts for brand voice analysis and SEO optimization, along with templates for content calendars and brand guidelines. Additionally, the copywriting skill is introduced to help generate conversion-focused marketing copy with strict anti-fabrication rules.

plugins/agentic-bundle-aas-localization-international-growth/skills · high confidence

New Agentic Skills for SaaS Launch and Revenue

Added a suite of new AI agent skills in the \plugins/agentic-bundle-aas-saas-launch-revenue/skills\ directory to guide users through SaaS product launches and revenue generation. The new skills include \analytics-product\ for product analytics and retention, \competitive-landscape\ for market positioning, \competitor-alternatives\ for SEO comparison pages, \copywriting\ for conversion-focused marketing, \email-sequence\ for nurture campaigns, \launch-strategy\ for phased go-to-market planning, \micro-saas-launcher\ for indie hacker MVPs, \monetization\ for Stripe integration and pricing, and \onboarding-cro\ for user activation. Each skill provides structured frameworks, code examples, and best practices for its specific domain.

plugins/agentic-bundle-aas-saas-launch-revenue/skills · high confidence

New Python skill scripts and templates added

This change introduces a collection of new Python scripts and templates across several skills in the Agentic Awesome Skills plugin. These include a production-ready FastAPI REST API template with security middleware, data models for a context agent, a read-only structural checker for Drizzle migrations, and Hugging Face vision trainer scripts for image classification, object detection, and SAM segmentation. Additionally, it adds a blueprint linter for structural validation, a base scraper for Brazilian auctioneers, and data schemas for a last-30-days research report skill.

python · high confidence

New QA and testing skill library added

The \plugins/agentic-bundle-aas-qa-test-automation/skills\ directory now includes a comprehensive set of documentation-based skills for quality assurance and test automation. These new files provide guidance and patterns for API contract testing, browser automation (including Playwright), CI/CD pipeline setup, code review checklists, end-to-end (E2E) testing patterns, and k6 load testing. This addition equips the system with structured instructions for verifying API compatibility, automating UI interactions, establishing quality gates, and validating performance.

plugins/agentic-bundle-aas-qa-test-automation/skills · high confidence

New accessibility and browser automation skills added

Added a suite of new agent skills to the skills directory to enhance accessibility auditing and browser automation capabilities. This includes \accesslint-audit\ for WCAG 2.2 compliance reporting and fixing, \accesslint-diff\ for comparing accessibility violations between revisions, and \accesslint-scan\ for live page auditing. Additionally, \playwright-skill\ provides a comprehensive framework for browser automation with helper utilities, while \fixing-accessibility\ and \ui-a11y\ offer guidelines for correcting HTML accessibility issues like ARIA labels and keyboard navigation. The bundle also introduces \screen-reader-testing\ for validating assistive technology compatibility and \ui-states\ for handling edge-case UI scenarios such as empty, loading, and error states.

plugins/agentic-bundle-aas-accessibility-inclusive-ux/skills · high confidence

New architecture design skills for DDD, CQRS, and event sourcing

Added a suite of new skills in the architecture-design plugin to guide complex system design, including Domain-Driven Design (strategic and tactical), CQRS implementation, event sourcing, and microservices patterns. These skills provide structured instructions, templates, and implementation playbooks for tasks such as defining bounded contexts, mapping context relationships, designing event stores, building projections, and managing distributed system resilience.

plugins/agentic-bundle-architecture-design/skills · high confidence

New curated automation and integration skills added to the bundle

The skills directory now includes new documentation for several automation capabilities: Airtable, GitHub, Google Calendar, Google Sheets, and HubSpot CRM operations via Rube MCP; browser automation with Playwright best practices; API integration design patterns (webhooks, event-driven architecture, sagas); and API rate-limit handling with retry/backoff strategies. These additions expand the curated bundle compositions available to users for programmatic workflow automation.

plugins/agentic-bundle-aas-automation-builder/skills · high confidence

New data layer for workflow catalog, SEO landing pages, and specialized plugins

The web app now includes a structured data layer in \apps/web-app/src/data\ that powers the new workflow explorer and SEO landing pages. This change introduces curated workflow definitions (\workflows.ts\) with search and skill-resolution capabilities, SEO landing page logic (\seoLandingPages.ts\) that curates skills based on editorial picks and topic scoring, and specialized plugin definitions (\specializedPlugins.ts\) derived from existing bundle manifests. Supporting JSON files (\docs.json\, \docs-social.json\, \seoLandingPages.json\) provide the static content for these features, and comprehensive tests (\workflows.test.ts\, \seoLandingPages.test.ts\) ensure the catalog adapter and SEO curation logic work correctly. Also in this area: 3 commits (0 fixes) modifying apps/web-app.

apps/web-app/src/data · medium confidence · unverified

New indie game development skills added

Added new skill definitions for 2D and 3D game development, game design, game audio, algorithmic art, and Godot GDScript patterns. These skills provide principles and best practices for sprite systems, tilemaps, physics, camera systems, rendering pipelines, shaders, level-of-detail, adaptive audio, core loops, and state machines.

plugins/agentic-bundle-indie-game-dev/skills · high confidence

New lint-and-validate skill for automated code quality checks

A new 'lint-and-validate' skill has been added to the agentic bundle, providing a structured procedure and bundled Python helpers to run configured lint and type checks for Node.js and Python projects. The skill introduces \lint\_runner.py\ to detect project types and execute local linters (such as ESLint, TSC, Ruff, and MyPy) with strict safety guards against silent network fetches or unauthorized fixes, and \type\_coverage.py\ to generate a bounded, read-only inventory of type annotations without inferring pass/fail verdicts. This enables agents to distinguish between actual lint failures and checks that did not run, ensuring that validation results are reported concretely and safely.

plugins/agentic-bundle-skill-author/skills/lint-and-validate · high confidence

New lint-and-validate skill with bundled runner scripts

A new 'Lint and Validate' skill has been added to the Agentic Awesome Skills bundle, providing a structured procedure for running configured lint and type checks on Node.js and Python projects. This entry includes the skill definition (SKILL.md) and two new helper scripts: \lint\_runner.py\, which detects project types and executes local linters (such as ESLint, TSC, Ruff, and MyPy) with bounded output and timeout handling, and \type\_coverage.py\, which performs a read-only, bounded inventory of type annotations in Python and TypeScript files without inferring pass/fail status. Users can now invoke this skill to distinguish actual lint failures from checks that did not run, ensuring more reliable validation results during development.

plugins/agentic-bundle-essentials/skills/lint-and-validate · high confidence

New marketing and SEO agent skills added

Added new agent skills for A/B test setup, analytics tracking, content creation, and content strategy. The A/B test setup skill provides a structured workflow for designing experiments, including hypothesis locking, sample size estimation, and tracking verification. The analytics tracking skill offers a framework for auditing and improving analytics implementation with a measurement readiness index. The content creator skill includes templates and local Python scripts for brand voice analysis and SEO optimization, while the content strategy skill provides guidance on topic clustering, keyword research, and content types.

plugins/agentic-bundle-aas-marketing-seo-growth/skills · high confidence

New repository tooling scripts for catalog management and security hardening

This change introduces a suite of new utility scripts in the \tools/scripts\ directory to improve catalog integrity, security, and maintenance automation. Key additions include \\_safe\_files.py\ and \build-aas-v1-offline-catalog.js\ to enforce strict file trust boundaries by rejecting symlinks and ensuring deterministic, signed catalog archives. The \apply\_skill\_optimization.cjs\ script isolates the application of skill optimizations to trusted review comments, preventing execution from arbitrary issue comments. Additionally, \audit\_consistency.py\ and \audit\_search\_migration\_readiness.js\ provide automated checks for metadata synchronization and search migration readiness, while \auto\_categorize\_skills.py\ and \analyze\_voltagent\_repo.py\ assist in maintaining and expanding the skill catalog with intelligent categorization and upstream analysis.

tools/scripts · high confidence

New scripts for skill activation and documentation validation

Added \activate-skills.bat\ and \activate-skills.sh\ to manage the Antigravity skills lifecycle, including syncing the skills library from the repository, merging from archives, expanding bundles via a Python helper (with a safe fallback for essentials), and restoring selected skills while validating input IDs. Also added \validate-links.sh\ to scan Markdown files in \README.md\, \docs\, and \docs\_zh-CN\ for broken internal links, and \validate-glossary.sh\ to check the \docs\_zh-CN/.glossary.json\ structure, term counts, and translation completeness.

scripts · high confidence

New security and architectural skills added to the Agentic Awesome Skills plugin

The plugin now includes two new agent skills: '00-andruia-consultant', which acts as a solutions architect to diagnose project status and create technical roadmaps in Spanish, and '007', a comprehensive security audit and hardening skill. The '007' skill provides detailed guidance for threat modeling (STRIDE/PASTA), OWASP compliance, incident response playbooks, and API security patterns, supported by reference documents and Python configuration scripts for automated scanning and scoring.

plugins/agentic-awesome-skills/skills · high confidence

New security and maintenance tooling libraries for the repository

This change introduces a suite of new libraries in tools/lib to harden repository security and automate maintenance workflows. It adds distribution artifact verification (distribution-artifact.js) to validate provenance and checksums, and strict file-system safety checks (symlink-safety.js, skill-utils.js) to prevent symlink attacks and unsafe path traversal. Maintenance operations are now governed by a state manager (maintenance-state.js) that uses cryptographic review receipts and file locking, alongside impact classification (maintenance-impact.js) to determine the necessary validation scope for changes. The update also includes a PR decision engine (pr-decision.js) to route changes based on risk, a reviewed fork ledger (reviewed-fork-skills.js) to track approved external contributions, and utilities for parsing raw Git diffs (git-raw-diff.js, git-blob-sizes.js) and filtering skills to manage token limits (skill-filter.js). Also in this area: 5 commits (1 fix) modifying tools.

tools/lib · medium confidence · unverified

New security testing and auditing skills added to the Agentic Awesome Skills bundle

The \plugins/agentic-bundle-aas-security-engineer/skills\ directory now includes a comprehensive set of new security-focused skills: \api-security-testing\ for REST and GraphQL security workflows, \burp-suite-testing\ for web application penetration testing with mandatory authorization gates, \cloud-penetration-testing\ for Azure, AWS, and GCP assessments, \ethical-hacking-methodology\ covering the full pentest lifecycle, \linux-privilege-escalation\ for identifying and exploiting Linux misconfigurations, \sast-configuration\ for setting up static analysis tools like Semgrep and SonarQube, and \security-auditor\ for DevSecOps and compliance auditing. These skills provide structured workflows, reference scripts, and safety constraints for authorized security testing and assessment activities.

plugins/agentic-bundle-aas-security-engineer/skills · high confidence

New skill discovery, comparison, and installation preview tools

The web app now includes several new components to help users find, evaluate, and prepare to install skills. The Outcome Explorer allows users to describe a goal to find relevant skills, while the Workflow Explorer provides curated, multi-phase execution paths. Users can compare selected skills in the Shortlist Review, which generates a copyable agent brief and an installation preview command via the Installation Handoff component. Additional components include a local-only Selection Feedback form, a Docs Search with result highlighting, a Docs Code Block with copy functionality, and updated Skill Cards with a local save button.

apps/web-app/src/components · high confidence

New systematic debugging skill and supporting utilities

The systematic-debugging skill has been added to the agentic-bundle-essentials plugin, introducing a four-phase process (Investigation, Pattern Analysis, Hypothesis, Implementation) that mandates root-cause analysis before any fix. The skill includes guidance on defense-in-depth validation, root-cause tracing, and condition-based waiting to replace flaky arbitrary timeouts. Supporting files include a TypeScript implementation for condition-based waiting (\condition-based-waiting-example.ts\), a bash diagnostic script to identify test polluters (\find-polluter.sh\), and several illustrative test scenarios to validate the skill's behavior under pressure.

plugins/agentic-bundle-essentials/skills/systematic-debugging · high confidence

New writing-skills reference library for agent skill authoring

Added a comprehensive set of documentation and templates in the \writing-skills\ directory to guide the creation of high-quality agent skills. This includes a main dispatcher (\SKILL.md\) with a decision tree for skill tiers, detailed best-practice guides covering conciseness, structure, and persuasion principles, and specific references for Claude Search Optimization (CSO), anti-rationalization techniques, and testing methodologies. The update also provides copy-paste templates for four skill categories (Technique, Reference, Discipline, Pattern) and a Graphviz convention file for process diagrams, establishing standardized naming, metadata, and directory structures for skill development.

plugins/agentic-bundle-skill-author/skills/writing-skills · high confidence

Redesigned web-app pages with new landing, docs, and plugin interfaces

The web application introduces a complete visual and structural overhaul of its core pages. The new Landing page features a neon-themed hero section, an inline introduction video, and clear entry points to the AAS Core catalog and Workbench. The Documentation page has been rebuilt with a sticky sidebar, client-side search, and an 'On this page' outline for easier navigation. A new Plugins page allows users to browse and filter specialized domain-specific distributions. The Home page now supports advanced catalog filtering (by risk, source, category) with URL-synced state for shareable links. Additionally, a dedicated 404 page with SEO metadata and a test suite have been added to handle missing routes gracefully.

apps/web-app/src/pages · high confidence

Security

Enforce read-only access on skill\_stars table via Row Level Security

A new migration locks the \skill\_stars\ table to read-only for anonymous and authenticated users by enabling Row Level Security and revoking write privileges. This change prevents the public browser key from performing \POST\, \PATCH\, or \DELETE\ operations, addressing coordinated security findings. The migration also revokes privileges on all public sequences to prevent potential write access through sequence manipulation.

supabase · high confidence

Establishes maintainer security and operational guidelines

The repository now enforces strict security and operational boundaries for maintainers. Local secrets are isolated in a gitignored \.env.local\ file, and a new \AGENTS.md\ mandates the use of the \antigravity-maintainer-batch-release\ skill for all maintenance and release workflows. Additionally, \.snyk\ is introduced to address security advisories, and \.vercelignore\ is configured to exclude sensitive build artifacts from deployment.

(repo-wide) · high confidence

Behavioural changes

1 commit (0 fixes) modifying (repo-wide)

A change to existing behaviour in (repo-wide) — 1 commit, 62 files: “feat: add reviewed HyperFrames video and media toolkit (\#2227)” (refs \#2227).

(repo-wide), apps/core-remote · medium confidence · unverified

2 commits (0 fixes) modifying skills/media-use/audio/scripts, skills/media-use/scripts

A change to existing behaviour in skills/media-use/audio/scripts, skills/media-use/scripts — 2 commits, 45 files: “feat: add reviewed HyperFrames video and media toolkit (\#2227)”; “feat: refresh seven HyperFrames skills to v0.9.15 (\#2281)” (refs \#2227, \#2281).

skills/media-use/audio/scripts, skills/media-use/scripts · medium confidence · unverified

20 commits (0 fixes) modifying plugins/agentic-bundle-design-studio

A change to existing behaviour in plugins/agentic-bundle-design-studio — 20 commits, 155 files: “chore: release v19.12.1 (\#2019)”; “\[skip pages\] chore: synchronize canonical repository state”; “chore: release v20.4.0 (\#2240)”; and 14 more (refs \#2019, \#2240, \#1998, \#2016, \#2050, …).

(repo-wide) · medium confidence · unverified

3 commits (1 fix) modifying (repo-wide)

A change to existing behaviour in (repo-wide) — 3 commits (1 fix), 852 files: “fix: refresh the official Codex marketplace manifest sources (\#2321)”; “chore: release v20.9.0 (\#2345)”; “chore: release v20.8.0 (\#2327)” (refs \#2321, \#2345, \#2327).

(repo-wide) · medium confidence · unverified

4 commits (1 fix) modifying repository

A change to existing behaviour in repository — 4 commits (1 fix), 3 files: “fix: complete individual skill review batch 16 (150 trees) (\#2096)”; “chore: release v20.0.0 (\#2106)”; “chore: Rename project to Agentic Awesome Skills”; and 1 more (refs \#2096, \#2106, \#243).

repository · medium confidence · unverified

4 commits (1 fix) modifying tools/lib/aas-v1/stack

A change to existing behaviour in tools/lib/aas-v1/stack — 4 commits (1 fix), 3 files: “feat: add AAS agent-first control plane preview (\#883)”; “feat: make Core selection agent-owned (\#934)”; “fix: expose safe Core validation details (\#938)”; and 1 more (refs \#883, \#934, \#938, \#1978).

tools/lib/aas-v1/stack · medium confidence · unverified

Introduce deterministic transaction journaling and filesystem safety checks

The AAS v1 transaction subsystem now enforces strict determinism and security for transaction logs and file operations. A new journaling mechanism (journal.js) records transaction events with cryptographic digests, enabling recovery from interrupted operations and detecting data corruption or 'torn writes'. Filesystem interactions are hardened against unsafe conditions: symbolic links are rejected, directory traversal is validated to prevent path escapes, and file ownership is verified. Additionally, a new digest module (digest.js) provides deterministic SHA-256 hashing for file trees, ensuring consistent state verification across runs.

tools/lib/aas-v1/transaction · high confidence

Introduces hardened, immutable catalog and runtime caching

The \tools/lib/aas-v1/cache\ module now provides a secure, immutable caching layer for catalog releases and runtime packages. It enforces strict integrity verification using SRI hashes and SHA-256 digests, validates archive contents against allowlists, and applies strict limits on file counts, sizes, and compression ratios to prevent resource exhaustion. The implementation hardens against path traversal, unsafe permissions, and symbolic link attacks by validating paths, rejecting special files, and using crash-safe atomic promotions via temporary staging directories.

tools/lib/aas-v1/cache · high confidence

New CLI installation handoff and safety validation logic

The CLI now includes a dedicated \installation-handoff\ module and enhanced safety checks in \main.js\ to manage the transition from selection to installation. This introduces strict validation for installation destinations (preventing invalid characters and path traversal) and skill IDs, ensuring that only authorized, well-formed selections are passed to the underlying npm execution. The change also adds robust file-system safety measures, such as atomic writes with integrity verification and symlink protection, to prevent corruption or security issues during the release and installation process.

tools/lib/aas-v1/cli · high confidence

Restore licensed upstream helpers and runtime support

This change restores previously removed helper scripts and runtime support for the docs-generator and frontend-slides skills. For the docs-generator, it adds PowerShell utilities to reliably resolve the correct PowerShell host executable (handling both Windows PowerShell 5.1 and PowerShell 7+) and to safely resolve the project root directory. For the frontend-slides skill, it introduces a confined Node.js server that serves deck files from a specified directory, ensuring strict path validation, limiting directory depth and file count, and blocking access to symlinks or files outside the designated deck root to prevent path traversal attacks.

skills/docs-generator/scripts, skills/frontend-slides-frontend-slides/scripts · high confidence

Updated AAS v1 catalog to version 18.13.0

The AAS v1 data catalog has been updated to package version 18.13.0. This update refreshes the asset manifest, including the skill content index and ontology files, and increases the total skill count to 2,638 assets.

data/aas-v1 · high confidence

Updated public assets and SEO metadata for the web application

This change updates the public-facing assets and metadata for the web application. It includes new or refreshed branding images (wordmarks, logos, favicons, and social cards), updated SEO files (sitemap, robots.txt, manifest), and refreshed LLMs metadata and skill counts. It also adds Google Search Console verification files and a backup skills catalog file.

apps/web-app/public · medium confidence · unverified

Web app restructured with new layout, dark theme, and lazy-loaded routes

The web application's source code has been reorganized to support a redesigned user interface and improved performance. The main entry point (App.tsx) now implements a sticky header with a new navigation structure (Home, Core, Workbench, Plugins, Docs) and integrates Vercel Web Analytics. Styling has been consolidated into index.css, establishing a dark-mode-only design system with CSS variables for colors, typography (Outfit and JetBrains Mono), and spacing, while replacing legacy Vite starter styles. Additionally, page components are now lazy-loaded via React Suspense to optimize initial load times, and the app is configured to render within a StrictMode React root.

apps/web-app/src · high confidence

Fixes

1 commit (1 fix) fixing .githooks

A fix in .githooks — 1 commit (1 fix), 1 file: “fix: streamline CI evidence, registry previews and cache persistence (\#2237)” (refs \#2237).

.githooks · medium confidence · unverified

1 commit (1 fix) fixing apps/web-app/src/vendor

A fix in apps/web-app/src/vendor — 1 commit (1 fix), 23 files: “fix(web): render Stageshow demos without iframes (\#2267)” (refs \#2267).

apps/web-app/src/vendor · medium confidence · unverified

2 commits (2 fixes) fixing skills/loki-mode/examples/todo-app-generated/frontend

A fix in skills/loki-mode/examples/todo-app-generated/frontend — 2 commits (2 fixs), 12 files: “Fix: Ensure all skills are tracked as files, not submodules”; “fix: complete individual skill review batch 17 (118 trees) (\#2097)” (refs \#2097).

skills/loki-mode/examples/todo-app-generated/frontend · medium confidence · unverified

Harden vercel-optimize sanitizers and add budget scope controls

The vercel-optimize skill now includes stricter validation and sanitization logic to prevent incorrect caching and improve audit clarity. Auth-like routes (matching patterns such as login, logout, auth, account, dashboard, checkout, cart, profile, session, and me) are explicitly disqualified from edge caching to prevent user state exposure. A new budget summary system introduces a configurable candidate limit (defaulting to 6) that gates the depth of investigation; when the default budget skips candidates, the tool prompts the user to expand the scope rather than silently ignoring issues. Additionally, citation sanitization ensures that only documentation references applicable to the user's specific framework and version are included in recommendations, and cost coverage reporting now explicitly identifies uncovered billing dimensions (such as Sandbox, AI Gateway, and Build Minutes) as blind spots in the audit.

skills/vercel-optimize · high confidence

Reproducible SEO metadata and social cards for documentation

The web-app build now generates stable, reproducible SEO artifacts for documentation routes. Sitemap lastmod dates are derived from Git history rather than the build wall-clock, and canonical-sync commits are excluded so dates remain byte-stable across syncs. A new social card template renders consistent preview images for docs, and verification scripts enforce correct metadata, link integrity, and asset provenance. These changes improve search discovery reliability and reduce build-time drift for documentation surfaces.

apps/web-app/scripts · high confidence

Robust skills loading with fallbacks and validation

The web app now uses a new SkillContext to load the skills catalog with improved reliability. It attempts to fetch the skills index from multiple candidate URLs to handle different deployment paths (such as GitHub Pages), validates that the response is a proper JSON array, and ignores stale or failed requests. Star counts are fetched asynchronously so they do not block the main skills data, ensuring the catalog remains available even if the star service is slow or unavailable.

apps/web-app/src/context · high confidence

Test coverage

Added test coverage for web-app pages and components; Added test coverage for web-app utility modules; Added test environment setup with browser API mocks; Added tests for SkillContext loading and validation logic; Added tests for landing page, route lazy loading, workbench isolation, and refresh-skills security; New test coverage for AAS v1 core, CLI, adapters, and security boundaries.

Dependencies

Introduce AAS Plugin Server and update web-app dependencies

The repository now includes a new \apps/aas-plugin\ package (\aas-plugin-server\) that bundles the \@modelcontextprotocol/sdk\ (v1.31.0) and \esbuild\ (v0.28.2) to support the local Codex plugin. The \web-app\ has been updated to use React 19.2.8, React DOM 19.3.0, React Router 8.3.0, and Vite 8.0.16, with the Node engine requirement expanded to support versions 22, 24, and 26. Additionally, the root package now bundles \yaml\ (v2.9.0) and \fast-uri\ (v3.1.8) to ensure runtime availability for CLI tools.

(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 45 → 41 (-4.0)
  • Rubric changed (rubric-2026.11.1 → rubric-2026.11.4) — scores are not directly comparable.

Lenses

  • Code Health 44 → 31 (-13.6)
  • Architecture 87 → 84 (-2.2)
  • Maturity 84 → 94 (+9.5)
  • Readiness 40 → 39 (-1.1)
  • Security 72 → 69 (-2.6)
  • Accessibility 39 → 47 (+7.3)
  • Performance 60 → 60 (+0.0)

Resolved (1197)

  • Classify ADR enforcement: Architecture Decision Records (skills/architecture-decision-records/SKILL.md)
  • Classify ADR enforcement: RFC 0001: The Override Protocol (skills/product-design-harness/resources/docs/rfcs/0001-override-protocol.md)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/cline-pilot/scripts/session_report.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/hugo-to-markdown/scripts/inventory_hugo_rules.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/landing-page-generator/scripts/landing_page_scaffolder.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/geo_audit.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/geo_audit.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/geo_audit.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/launch_check.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/launch_check.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/polish_audit.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/launch-check/scripts/polish_audit.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/liuguang-banlan-ui/scripts/validate_manifest.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/matplotlib/scripts/plot_template.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/matplotlib/scripts/style_configurator.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/podcast-generation/scripts/pcm_to_wav.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/slideops/scripts/cite.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/stability-ai/scripts/generate.py)
  • Duplicated block (10 lines × 3) (plugins/agentic-awesome-skills-claude/skills/usage-engineer/scripts/ue.py)
  • Duplicated block (10 lines × 4) (plugins/agentic-awesome-skills-claude/skills/app-store-optimization/aso_scorer.py)
  • …and 1177 more

New (695)

  • Changelog.Changelog (cognitive 20) (apps/web-app/src/pages/Changelog.tsx)
  • Changelog.Changelog (cyclomatic 21) (apps/web-app/src/pages/Changelog.tsx)
  • CoreOverview.CoreOverview (cognitive 17) (apps/web-app/src/pages/CoreOverview.tsx)
  • CoreRecordedExample.CoreRecordedExample (cognitive 17) (apps/web-app/src/pages/CoreRecordedExample.tsx)
  • CoreRecordedExample.CoreRecordedExample (cyclomatic 16) (apps/web-app/src/pages/CoreRecordedExample.tsx)
  • DesignSystemGenerator._select_best_match (cognitive 18) (plugins/aasb-design-studio/skills/ui-ux-pro-max/scripts/design_system.py)
  • Duplicated block (10 lines × 10) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/metadata_optimizer.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/ab_test_planner.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/aso_scorer.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/aso_scorer.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/aso_scorer.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/review_analyzer.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-design-studio/skills/mobile-design/scripts/mobile_audit.py)
  • Duplicated block (10 lines × 5) (plugins/aasb-design-studio/skills/mobile-design/scripts/mobile_audit.py)
  • Duplicated block (10 lines × 7) (plugins/aasb-aas-accessibility-inclusive-ux/skills/webapp-testing/scripts/with_server.py)
  • Duplicated block (10 lines × 7) (plugins/aasb-aas-api-platform-builder/skills/api-design-principles/assets/rest-api-template.py)
  • Duplicated block (10 lines × 8) (plugins/aasb-design-studio/skills/ui-ux-pro-max/scripts/design_system.py)
  • Duplicated block (11 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/competitor_analyzer.py)
  • Duplicated block (11 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/competitor_analyzer.py)
  • Duplicated block (11 lines × 5) (plugins/aasb-aas-mobile-app-builder/skills/app-store-optimization/keyword_analyzer.py)
  • …and 675 more

Changes since last survey

  • 68 commits — 49 feature/other, 19 fixes

By area

  • apps/web-app — 13 commits
  • (root) — 9 commits
  • tools/scripts — 4 commits
  • docs/users — 2 commits
  • plugins/agentic-awesome-skills — 2 commits
  • plugins/agentic-awesome-skills-claude — 2 commits
  • skills/stageshow — 2 commits
  • skills/teardown — 2 commits
  • apps/aas-plugin — 1 commit
  • apps/core-remote — 1 commit
  • docs/maintainers — 1 commit
  • plugins/aasb-design-studio — 1 commit
  • skills/accesslint-diff — 1 commit
  • skills/ai-code-review-triage — 1 commit
  • skills/antigravity-maintainer-batch-release — 1 commit
  • skills/aomi-transact — 1 commit
  • skills/autorekt — 1 commit
  • skills/bowmark — 1 commit
  • skills/career-ops — 1 commit
  • skills/cloro-mcp — 1 commit

Notable commits

  • fix: fix(accesslint-diff): remove angle brackets from the description (#2300)
  • fix: fix(stageshow): update playground dependencies and document runtime support (#2264)
  • fix: fix(web): render Stageshow demos without iframes (#2267)
  • fix: fix(web): restore embedded demos and simplify navigation (#2263)
  • fix: fix: bind copied skill review receipts to destination trees (#2305)
  • fix: fix: clarify AAS identity and skill discovery (#2311)
  • fix: fix: correct VoiceMoat connector details in 8 skills (#2314)
  • fix: fix: describe the own NetAPI IP feed in the netapi skill (#2272)
  • fix: fix: enforce the complete release notes format (#2270)
  • fix: fix: harden runtime, installer and development asset boundaries (#2295)
  • fix: fix: harden viewer checks and finalize catalog maintenance (#2307)
  • fix: fix: isolate Codex marketplace bundles from portable manifests (#2319)
  • fix: fix: make discovery pages crawlable and strengthen SEO checks (#2310)
  • fix: fix: migrate Aomi CLI and MCP URLs to aomi.chat (#2332)
  • fix: fix: omit unverified guide dates in hosted builds (#2312)
  • fix: fix: preserve installer backups and improve catalog discovery (#2309)
  • fix: fix: refresh career-ops skill wording from upstream (#2255)
  • fix: fix: refresh the official Codex marketplace manifest sources (#2321)
  • fix: fix: sync pixeltable skill with pixeltable-skill 9296e22 (#2248)
  • change: Add CrakSEO MCP connect-triggering agent skills (#2298)
  • …and 48 more

Architecture

  • Unchanged — 0 containers · 2 contexts · 0 edges

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

Survey your own repository

sickn33/agentic-awesome-skills 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 15 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 8f1eb06978b026c5434284202c22215792c3071c — 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-2026.11.73+preprod-52de90b74216.