| title | Agent Constitution 📜 |
|---|---|
| description | Universal validated context architecture for AI Agents (VS Code, Antigravity, Claude) |
| location | README.md |
| last_updated | 2026-06-23 |
The Validated Context Architecture for AI Agents.
368 skills · 63 agents · 79 commands · Works with Claude Code, Cursor, Codex, Gemini CLI
bash <(curl -fsSL https://raw.githubusercontent.com/su6i/agent-constitution/main/install.sh)
Regenerate:
vhs assets/demo.tape
The main technical document of this repository is AGENTIC-CODING-SETUP.md.
If you want the core benchmark analysis, cost model, routing strategy, setup patterns, and low-cost agentic coding methodology, start there first. The rest of the repository either:
- introduces that document,
- operationalizes it,
- localizes it,
- or extends it with reusable rules, workflows, and templates.
For the repository map, see docs/INFORMATION-ARCHITECTURE.md.
Most AI Agents (Cursor, AntiGravity, Windsurf, Copilot) fail because their "memory" is unstructured. You give them a 50-page prompt, they hallucinate. You give them nothing, they write spaghetti code. We needed a middle ground: A strict, modular "Constitution" that forces Agents to behave like Senior Engineers.
This repository is not just "rules". It is a Modular Context Architecture. It breaks down the software lifecycle into 5 atomic, linked workflows. The Agent loads only what it needs, when it needs it.
In practice, the repository now has a clear center of gravity:
- AGENTIC-CODING-SETUP.md is the flagship guide and canonical technical reference.
- README.md is the onboarding and navigation layer.
- AGENTS.md is the execution contract and the canonical entry point for every agent harness.
GEMINI.md,GROK.md,QWEN.md,MINIMAX.md,.cursorrules,.windsurfrules,.clinerules, and.github/copilot-instructions.mdare thin bootloaders: whichever config file a tool reads natively, it lands there and is routed to AGENTS.md →rules/. No rules are duplicated in them (rule 045)..agent/,templates/,bin/, anddocs/are the implementation and support layers around that core guidance.
Every bootloader tells the agent to read rules/DIGEST.md
first — it is the auto-generated digest of every non-negotiable section in
rules/*.md (markers wrap the full original text, not a compressed
summary). Generated by bin/generate-digest.sh
from HTML-comment markers (<!-- digest:start --> … <!-- digest:end -->);
CI runs bin/generate-digest.sh --check on every push/PR and fails if the
digest drifts from the source rules (see
rules/045 §Digest Mechanism).
Reading the digest is attested with bin/ack-rules.sh: it
prints the digest to stdout (so it enters the agent's context) and writes a
gitignored .rules-ack holding the digest hash. Where
templates/hooks/pre-commit is installed, Rule 7
blocks a commit whose .rules-ack is missing or stale. It is hash-based only —
a new branch needs no fresh ack — it skips merge commits, and it is a silent
no-op in any repo that has not installed bin/ack-rules.sh.
The same hook's Rule 8 blocks a newly added or renamed path whose name
contains a non-ASCII byte (rules/000 §Language Policy).
Only new paths are checked, so it never fires on files that predate the rule,
and the file's content language is a separate question.
- ⚖️ The Neural Gavel: A strict
.cursorrulesrouter that prevents the Agent from guessing. - 🧠 Modular Memory: Workflows for Init, Docs, AI, and QA are split to prevent "Lost-in-the-Middle" errors.
- 🛡️ Truth Protocol: Agents are forbidden from marking tasks "Done" without
ls -Rverification. - 🤖 Anti-Hallucination: Strict file collision and deletion safety protocols.
- Read AGENTIC-CODING-SETUP.md for the core methodology.
- Read AGENTS.md for execution constraints and agent behavior.
- Use the workflow, template, and skill files as modular extensions of the core guide.
-
Install the Scaffolder:
# Add this alias to your shell config (~/.zshrc) alias init-gh='~/path/to/agent-constitution/bin/scaffold.sh'
-
Run in any new project:
mkdir my-new-project && cd my-new-project init-gh
Result: The
rules/,workflows/, and.agent/prompts/are injected and committed automatically.
- Clone this repo.
- Copy
rules/,workflows/, and.agent/prompts/to your project root. - Ask your Agent:
"Audit my codebase against the Quality Assurance protocol."
Before using this constitution in your own project, replace the author email with your own:
| File | Lines | What to change |
|---|---|---|
rules/040-git.md |
51, 52, 53, 81 | Replace <your-git-email> with your own email |
Why: The Commit Identity rule enforces a specific author email on every commit. The value shipped in this repo is the original author's address — it must be yours before you start using the rules.
# Quick replace (run from the repo root):
sed -i '' 's/<your-git-email>/your@email.com/g' rules/040-git.mdThen set your global git identity to match:
git config --global user.email "your@email.com"
git config --global user.name "Your Name"Three hooks turn the non-negotiable git rules (rules/040-git.md) into deterministic
gates that no agent or human can "forget". Their canonical sources live in
templates/hooks/ — installing them needs nothing but a copy into your repo's
.git/hooks/:
cp templates/hooks/pre-commit templates/hooks/pre-merge-commit templates/hooks/commit-msg /path/to/your-repo/.git/hooks/
chmod +x /path/to/your-repo/.git/hooks/pre-commit /path/to/your-repo/.git/hooks/pre-merge-commit /path/to/your-repo/.git/hooks/commit-msgIf you use the optional amir-cli automation,
amir init-project (new repos) and amir update-projects (existing repos)
install them for you.
| Hook | What it blocks |
|---|---|
pre-commit |
Direct commits to main/master (use a feature branch); the Docs Checklist — a code change must also touch a doc (README / CHANGELOG / docs/ / *.md); personal/memory files (TODO.md, SESSION.md, force-added CLAUDE.md, …) — deleting one is allowed, that's the remediation; secrets/PII in added lines; and a skills/**.md edit without a version: bump (rules/036-skill-versioning.md); plus rules attestation — a missing or stale .rules-ack (run bin/ack-rules.sh), hash-based, skipped on merges, and a no-op where bin/ack-rules.sh is not installed. |
pre-merge-commit |
The same privacy and skill-version gates applied to merge commits (git never runs pre-commit on automatic merges). Branch-protection and docs checks are skipped on merges — merging an approved branch into main is the sanctioned protocol step. |
commit-msg |
AI co-authorship — ever. Any Co-Authored-By: <AI> trailer, a "Generated with <AI>" line, or a 🤖 marker in the message. |
git commit --no-verify # bypass for ONE commit (leaves a shell-history trace)
rm .git/hooks/commit-msg # remove a single hook from this clone
rm .git/hooks/pre-commit
rm .git/hooks/pre-merge-commit
amir update-projects --no-hook # sync the constitution WITHOUT (re)installing hooksRemoving a hook only affects that one clone; the templates and the rules stay
intact, and a later amir update-projects reinstalls them unless you pass --no-hook.
- Agentic Coding 2026: Canonical benchmark, ROI, routing, tooling, and low-cost setup guide for the repository.
- Information Architecture: Explains how the rest of the repository relates to the flagship guide.
- Init Workflow: How to start clean.
- First Session: Onboard a scaffolded project — fill CLAUDE.md, verify structure.
- Propagate Updates: Roll new rules/skills/hook out to every project (
amir update-projects). - AI Logic: Architect vs Executor models.
- QA Protocol: Zero-bug policy.
- Communication: Standardized project reporting.
- Social Showcase: World-class marketing assets.
- Technical Template: Base structure for technical docs.
- LinkedIn Launch: Viral marketing hooks.
- Logo Specs: DALL-E/Midjourney prompts for tech branding.
Search this page (
Ctrl+F) by keyword — every skill name, tool, and technology is listed below.
AI & Machine Learning (26 skills)
| Skill | Description |
|---|---|
| ai-logic-patterns | Master prompting & agent orchestration rules |
| ai-regression-testing | Automated regression suites for LLM outputs |
| ai-router | Cost-aware model routing and fallback chains |
| ai-video-generation | Gen-video with Runway, Kling, Luma, Veo |
| agentic-engineering | Building robust agentic systems end-to-end |
| autonomous-loops | Self-driving agent loops with quality gates |
| autonomous-agent-harness | Production-ready autonomous agent scaffolding |
| continuous-learning | Agents that improve from session observations |
| continuous-learning-v2 | v2 learning pipeline with structured memory |
| eval-harness | LLM evaluation framework & scoring pipelines |
| fal-ai-media | fal.ai image & video generation API patterns |
| foundation-models-on-device | On-device inference (MLX, llama.cpp, Core ML) |
| llm-ml-workflow | Productionizing AI models end-to-end |
| llm-trading-agent-security | Security patterns for LLM-driven trading agents |
| ml-adoption-playbook | Organizational ML rollout & change management |
| multi-rag-orchestration | Stateful multi-step RAG with lexical tracking |
| prompt-engineering | Advanced system prompts, personas & few-shot |
| prompt-optimizer | Systematic prompt scoring and iteration |
| pytorch-patterns | PyTorch training loops, custom datasets, hooks |
| python-pytorch-sklearn | Sklearn pipelines with PyTorch models |
| r-lang-guide | targets pipelines & renv practices |
| ragas-evaluation | RAG evaluation with RAGAS metrics |
| regex-vs-llm-structured-text | When to use regex vs LLM for text extraction |
| reinforcement-learning | Gymnasium envs & Stable-Baselines3 training |
| recsys-pipeline-architect | Recommendation system pipeline design |
| token-budget-advisor | Context budget management & cost optimization |
Agent Systems & Orchestration (22 skills)
| Skill | Description |
|---|---|
| agent-architecture-audit | 12-layer agent stack diagnostic & audit |
| agent-eval | Agent performance evaluation frameworks |
| agent-harness-construction | Building production agent harnesses |
| agent-introspection-debugging | Debug agent behavior & silent failures |
| agent-self-evaluation | Agents that score their own outputs |
| agentic-os | Operating-system-level agent coordination |
| continuous-agent-loop | CI/PR-integrated continuous agent loops |
| context-budget | Token budget tracking and context hygiene |
| cost-tracking | Per-session and per-task cost instrumentation |
| cost-aware-llm-pipeline | Build pipelines that stay within cost targets |
| intent-driven-development | Acceptance criteria before implementation |
| moltbot-orchestration | Multi-agent video factory architecture |
| orch-pipeline | Full orchestration pipeline from plan to deploy |
| orch-build-mvp | Orchestrated MVP build workflow |
| orch-add-feature | Orchestrated feature addition workflow |
| orch-change-feature | Orchestrated feature change workflow |
| orch-fix-defect | Orchestrated defect fix workflow |
| orch-refine-code | Orchestrated code refinement workflow |
| parallel-execution-optimizer | Maximize parallelism in multi-agent runs |
| plan-orchestrate | Plan-then-orchestrate multi-step execution |
| recursive-decision-ledger | Decision audit trail for autonomous agents |
| team-agent-orchestration | Coordinate teams of specialized agents |
Video & Media Production (28 skills)
| Skill | Description |
|---|---|
| auto-editor | Automated silence removal & jump-cut editing |
| blender-motion-state-inspection | Debug Blender animation state machines |
| davinci-resolve-scripting | DaVinci Resolve Python API & timeline automation |
| fal-ai-media | fal.ai image/video generation (FLUX, Kling, Veo) |
| ffmpeg-recipes | Copy-pasteable FFmpeg commands for video automation |
| ffmpeg-reference | Codec ops, metadata, filters & flag reference |
| manim-video | Manim scene patterns & network graph explainers |
| motion-advanced | Advanced motion design principles & techniques |
| motion-foundations | Core motion design theory & easing |
| motion-patterns | Reusable motion design patterns |
| motion-ui | UI motion & micro-interaction design |
| obs-studio | OBS automation, scenes, and streaming setup |
| remotion-video-creation | Programmatic video with React + Remotion |
| taste | Creative direction layer for music videos & edits |
| video-blender-automation | Blender bpy API & Geometry Nodes scripting |
| video-editing | Professional video editing workflows & tools |
| video-effects-transitions | VFX, transitions & compositing techniques |
| video-manim-math | Manim math animations: OpenGL, plugins, shaders |
| video-production-automation | Full Python pipeline: Manim, MoviePy, OpenCV |
| video-remotion-react | React-based programmatic video creation |
| video-resolve-editing | DaVinci Resolve post-production & color grading |
| video-stick-figure | 2D stick figure animation & physics |
| videodb | VideoDB vector search & scene understanding |
| visual-ai-cinematography | Gen-video cinematography (Runway, Kling, Luma) |
| visual-character-consistency | Identity preservation across frames (ComfyUI, LoRA) |
| visual-director-procedural | Blender & Manim semantic visual direction |
| visual-thumbnail-psychology | CTR-optimized thumbnail design psychology |
| comfyui-stable-diffusion | ComfyUI workflows & Stable Diffusion nodes |
Voice, Audio & TTS (17 skills)
| Skill | Description |
|---|---|
| ai-dubbing-localization | Automated dubbing & tone-preserving localization |
| ai-sfx-generation | Latent diffusion for sound design & SFX |
| audio-processing | Neural denoising & EBU R128 normalization |
| fish-speech | Fish Speech TTS model setup & fine-tuning |
| gpt-sovits | GPT-SoVITS voice cloning & training |
| heygen-api | HeyGen avatar video generation API |
| huggingface-tts | HuggingFace TTS model catalog & inference |
| mlx-whisper | Apple Silicon Whisper transcription via mlx-whisper |
| music-generation | AI music generation (Suno, Udio, MusicGen) |
| opensource-tts | Open-source TTS stack comparison & setup |
| persian-tts-training | Persian-language TTS fine-tuning pipeline |
| storytelling-tts-m4-system | End-to-end TTS storytelling on Apple Silicon |
| subtitle-generator | Professional subtitles with cinematic typography |
| voice-ai-cloning-finetuning | Multilingual voice cloning & emotional synthesis |
| voice-dialogue-tts | Multi-speaker turn-taking & emotional prosody |
| voice-emotional-acting | One-person multi-character production |
| voice-orchestration-multi-model | Multi-model TTS pipelines (Fish, Dia, Bark, Parler) |
| voice-synthesis-multilingual | SOTA multilingual TTS & cross-lingual cloning |
| xtts-v2 | XTTS-v2 fine-tuning & streaming inference |
Content, YouTube & Marketing (21 skills)
| Skill | Description |
|---|---|
| article-writing | Long-form article structure & SEO writing |
| brand-discovery | Brand positioning & identity research |
| brand-voice | Consistent brand voice modeling & guidelines |
| competitive-platform-analysis | Systematic competitor platform teardown |
| competitive-report-structure | Competitive intelligence report templates |
| content-engine | Automated content production pipelines |
| copywriting | Conversion-focused copy & persuasion logic |
| crosspost | Cross-platform content distribution automation |
| marketing-campaign | Campaign planning, execution & measurement |
| screenwriting-automated | 100% automated script production pipeline |
| screenwriting-frameworks | 3-Act, Hero's Journey, beat sheets |
| screenwriting-youtube | High-retention YouTube hooks & psychology |
| seo | Technical SEO, on-page optimization & audits |
| social-graph-ranker | Social network analysis & influence ranking |
| social-publisher | Scheduled multi-platform social publishing |
| storytelling-clil-education | Educational storytelling & Leitner SRS |
| storytelling-narrative-frameworks | Save the Cat, Story Circle & advanced structures |
| x-api | X/Twitter API v2 patterns & automation |
| youtube-analytics | YouTube Data API analytics & reporting |
| youtube-automation-pipeline | End-to-end YouTube publishing automation |
| youtube-data-api | YouTube Data API v3 — uploads, playlists, captions |
| youtube-dlp-web-download | yt-dlp: Cloudflare bypass, stream selection |
| youtube-seo | Titles, thumbnails, retention & channel growth |
Research & Science (12 skills)
| Skill | Description |
|---|---|
| deep-research | Systematic multi-source research protocols |
| exa-search | Exa semantic search API integration |
| market-research | Primary & secondary market research methods |
| ml-adoption-playbook | Enterprise ML adoption & stakeholder alignment |
| prediction-market-oracle-research | Prediction market data sourcing & analysis |
| prediction-market-risk-review | Risk assessment for prediction market positions |
| research-ops | Research workflow automation & knowledge mgmt |
| scientific-db-pubmed-database | PubMed API queries & literature retrieval |
| scientific-db-uspto-database | USPTO patent search & analysis |
| scientific-pkg-gget | gget for genomics data retrieval |
| scientific-thinking-literature-review | Systematic literature review methodology |
| scientific-thinking-scholar-evaluation | Academic paper quality evaluation |
Python (8 skills)
| Skill | Description |
|---|---|
| python-core-standards | Project structure, uv, typing & conventions |
| python-containerization | Docker: Slim vs Alpine, multi-stage builds |
| python-github-setup | GitHub Actions, templates & semantic release |
| python-pandas-sklearn | Method chaining, pipelines & ColumnTransformer |
| python-patterns | Idiomatic Python patterns & anti-patterns |
| python-pytorch-sklearn | Sklearn data pipelines with PyTorch models |
| python-testing | pytest, fixtures, parametrize & coverage |
| generating-python-installer | Build distributable Python installer packages |
Web & Frontend (24 skills)
| Skill | Description |
|---|---|
| angular-developer | Angular architecture, RxJS & state management |
| bun-runtime | Bun runtime setup, bundling & testing |
| chrome-extension-best-practices | MV3 extensions, UI & Shadow DOM |
| design-system | Design tokens, component libraries & theming |
| fastapi-best-practices | FastAPI: Pydantic v2, lifespan, background tasks |
| fastapi-patterns | FastAPI project structure, auth, DI & testing |
| flask-json-guide | Robust Flask API structure & error handling |
| frontend-a11y | Web accessibility (WCAG, ARIA, screen readers) |
| frontend-design-direction | Visual direction & design review for UIs |
| frontend-patterns | Cross-framework frontend architecture patterns |
| frontend-slides | Browser-based presentation & slide tooling |
| js-ts-code-quality | Strict TypeScript, Biome/ESLint & Vitest |
| liquid-glass-design | iOS 26 liquid glass UI design patterns |
| make-interfaces-feel-better | Micro-interactions & polish for any UI |
| modern-web-ui | Vanilla HTML/CSS/JS best practices |
| nextjs-turbopack | Next.js App Router & Turbopack setup |
| nuxt4-patterns | Nuxt 4 architecture & composables |
| react-patterns | React component patterns & hooks |
| react-performance | React rendering optimization & profiling |
| react-testing | Testing Library, MSW & Vitest for React |
| ui-demo | Interactive demo & prototype patterns |
| vite-patterns | Vite config, plugins & build optimization |
| vue-patterns | Vue 3 Composition API & ecosystem patterns |
| ui-to-vue | Migrating UI components to Vue 3 |
Backend (40 skills)
| Skill | Description |
|---|---|
| backend-patterns | Cross-language backend architecture patterns |
| cpp-coding-standards | Modern C++20/23 standards & best practices |
| cpp-testing | Catch2, GoogleTest & CMake testing |
| csharp-testing | xUnit, Moq & .NET testing patterns |
| django-celery | Celery task queues with Django |
| django-patterns | Django project structure & ORM patterns |
| django-security | Django security hardening & OWASP |
| django-tdd | TDD with Django & pytest-django |
| django-verification | Django deployment verification checklists |
| dotnet-patterns | .NET 8+ architecture & dependency injection |
| error-handling | Structured error handling across languages |
| flask-json-guide | Flask JSON API patterns & error handling |
| fsharp-testing | F# testing with Expecto & FsCheck |
| golang-patterns | Go idioms, concurrency & project layout |
| golang-testing | Go testing: table tests, benchmarks, fuzz |
| hexagonal-architecture | Ports & adapters pattern implementation |
| java-coding-standards | Java 21+ records, sealed types & patterns |
| jpa-patterns | JPA/Hibernate patterns & query optimization |
| kotlin-coroutines-flows | Kotlin coroutines, Flows & structured concurrency |
| kotlin-exposed-patterns | Kotlin Exposed ORM patterns |
| kotlin-ktor-patterns | Ktor server patterns & plugin architecture |
| kotlin-patterns | Kotlin idiomatic patterns & conventions |
| kotlin-testing | Kotlin testing with Kotest & MockK |
| laravel-patterns | Laravel architecture & Eloquent patterns |
| laravel-plugin-discovery | Laravel package & plugin discovery |
| laravel-security | Laravel security & authorization patterns |
| laravel-tdd | TDD with Laravel & Pest |
| laravel-verification | Laravel deployment verification |
| nestjs-patterns | NestJS modules, providers & interceptors |
| perl-patterns | Modern Perl idioms & CPAN usage |
| perl-security | Perl security: taint mode & injection prevention |
| perl-testing | Perl testing with Test::More & prove |
| quarkus-patterns | Quarkus CDI, Panache & REST patterns |
| quarkus-security | Quarkus OIDC & security hardening |
| quarkus-tdd | TDD with Quarkus & QuarkusTest |
| quarkus-verification | Quarkus native build verification |
| rust-patterns | Rust ownership, traits & async patterns |
| rust-testing | Rust testing: unit, integration & cargo nextest |
| springboot-patterns | Spring Boot architecture & Spring Data |
| springboot-security | Spring Security & OAuth2 patterns |
| springboot-tdd | TDD with Spring Boot & Testcontainers |
| springboot-verification | Spring Boot production readiness checks |
| tinystruct-patterns | tinystruct Java framework patterns |
Mobile (12 skills)
| Skill | Description |
|---|---|
| android-clean-architecture | Android Clean Architecture & MVVM |
| compose-multiplatform-patterns | Kotlin Multiplatform + Compose patterns |
| dart-flutter-patterns | Flutter architecture & Dart patterns |
| flutter-dart-code-review | Flutter code review checklist |
| ios-icon-gen | iOS app icon generation & asset catalogs |
| jetpack-compose-guidelines | Android declarative UI & permissions |
| kotlin-patterns | Kotlin Android idioms & Coroutines |
| swift-actor-persistence | Swift actors & SwiftData persistence |
| swift-concurrency-6-2 | Swift 6.2 strict concurrency model |
| swift-protocol-di-testing | Swift protocol-based DI & testability |
| swiftui-guidelines | SwiftUI modern architecture & audio |
| swiftui-patterns | SwiftUI component patterns & state |
Database (6 skills)
| Skill | Description |
|---|---|
| clickhouse-io | ClickHouse OLAP queries & ingestion patterns |
| database-migrations | Safe schema migrations across databases |
| mysql-patterns | MySQL indexing, query optimization & replication |
| postgres-patterns | PostgreSQL: JSONB, CTEs, partitioning & RLS |
| prisma-patterns | Prisma ORM schema design & migrations |
| redis-patterns | Redis data structures, caching & pub/sub |
DevOps & Infrastructure (17 skills)
| Skill | Description |
|---|---|
| canary-watch | Canary deployment monitoring & rollback |
| config-gc | Garbage-collect stale config & dead flags |
| deployment-patterns | Blue/green, canary & rolling deploy patterns |
| docker-patterns | Dockerfile best practices & multi-stage builds |
| flox-environments | Flox reproducible dev environments |
| git-workflow | Git branching, rebase & PR workflow standards |
| github-code-quality | GitHub Actions CI, PR templates & code owners |
| github-ops | GitHub API, releases & repository automation |
| kubernetes-docs | K8s best practices & MkDocs integration |
| kubernetes-patterns | K8s manifests, Helm charts & operator patterns |
| latency-critical-systems | Sub-millisecond latency design & profiling |
| linux-cuda-python | HPC setup, PyTorch CUDA & GPU profiling |
| ops-automation | CI/CD, Docker & experiment tracking automation |
| production-audit | Pre-launch production readiness audit |
| python-containerization | Python Docker: Slim vs Alpine, multi-stage |
| uncloud | Cloud cost reduction & simplification patterns |
| data-throughput-accelerator | High-throughput data pipeline optimization |
Network & NetDevOps (10 skills)
| Skill | Description |
|---|---|
| cisco-ios-patterns | Cisco IOS/IOS-XE config & automation patterns |
| homelab-network-readiness | Homelab pre-deployment network checklist |
| homelab-network-setup | Homelab network design & VLAN architecture |
| homelab-pihole-dns | Pi-hole DNS-level ad blocking setup |
| homelab-vlan-segmentation | VLAN design & 802.1Q trunk configuration |
| homelab-wireguard-vpn | WireGuard VPN setup & peer management |
| netmiko-ssh-automation | Netmiko SSH automation for network devices |
| network-bgp-diagnostics | BGP troubleshooting & route analysis |
| network-config-validation | Network config diff & compliance validation |
| network-interface-health | Interface health monitoring & alerting |
Web3 & Blockchain (7 skills)
| Skill | Description |
|---|---|
| defi-amm-security | AMM security: reentrancy, flash loans, MEV |
| evm-token-decimals | EVM token decimal handling & precision |
| llm-trading-agent-security | Security for LLM-driven on-chain agents |
| nodejs-keccak256 | keccak256 hashing in Node.js |
| web3-react-dapps | Wagmi, Viem & dApp architecture |
| web3-solidity-foundry | Foundry: Rust-based Solidity testing & fuzzing |
| web3-solidity-hardhat | Hardhat JS/TS ecosystem & tooling |
Testing & QA (10 skills)
| Skill | Description |
|---|---|
| ai-regression-testing | Automated regression for LLM outputs |
| benchmark | Performance benchmarking methodology |
| benchmark-methodology | Rigorous benchmark design & statistical analysis |
| benchmark-optimization-loop | Iterative benchmark-driven optimization |
| browser-qa | Browser automation & visual regression testing |
| e2e-testing | End-to-end testing with Playwright & Cypress |
| ragas-evaluation | RAG pipeline evaluation with RAGAS |
| tdd-workflow | TDD: RED/GREEN/refactor + evidence reports |
| verification-loop | Automated verification gate patterns |
| windows-desktop-e2e | Windows desktop app E2E testing |
Security (9 skills)
| Skill | Description |
|---|---|
| django-security | Django security hardening & OWASP top 10 |
| gateguard | Pre-edit investigation gate for critical files |
| healthcare-phi-compliance | HIPAA PHI handling & de-identification |
| hipaa-compliance | HIPAA technical & administrative safeguards |
| laravel-security | Laravel security & XSS/CSRF prevention |
| safety-guard | Agent output safety filtering & guardrails |
| security-bounty-hunter | Bug bounty methodology & vulnerability research |
| security-review | Code security review checklist & SAST |
| security-scan | Automated dependency & secret scanning |
Claude Code & Agent Harness (18 skills)
| Skill | Description |
|---|---|
| agent-sort | Agent task routing & priority sorting |
| canary-watch | Deployment canary monitoring |
| claude-code-integration | Claude Code CLI, hooks, MCP & agentic patterns |
| claude-devfleet | Claude Code fleet management & parallelism |
| codehealth-mcp | MCP-driven code health metrics & alerts |
| configure-ecc | ECC harness configuration & setup |
| context-budget | Token budget tracking & context hygiene |
| cost-tracking | Per-session LLM cost instrumentation |
| dynamic-workflow-mode | Switch agent workflow modes at runtime |
| ecc-guide | ECC harness user guide & patterns |
| ecc-tools-cost-audit | Audit ECC tool usage & cost attribution |
| hookify-rules | Claude Code hook authoring & management |
| nanoclaw-repl | REPL-based agent session persistence |
| repo-scan | Repository structure & health scanning |
| rules-distill | Distill project rules from code patterns |
| skill-comply | Enforce skill usage compliance in agent runs |
| skill-scout | Discover missing skills from agent behavior |
| skill-stocktake | Audit & deduplicate skill catalog |
| token-budget-advisor | Context budget management strategies |
Scripting & Automation (10 skills)
| Skill | Description |
|---|---|
| automation-audit-ops | Audit & optimize automation pipelines |
| rtl-persian-app-patching | Force Persian/RTL into Electron & web apps you don't own |
| cli-table-alignment | Pixel-perfect ASCII tables with emoji support |
| dmux-workflows | tmux/dmux session & pane automation |
| hookify-rules | Claude Code hooks authoring & lifecycle |
| macos-automation | macOS Python/Zsh desktop automation scripts |
| terminal-ops | Terminal productivity & shell tooling |
| zsh-completion | Robust zsh autocomplete without pitfalls |
| zsh-scripting-advanced | SIGINT traps, option parsing & shell patterns |
| generating-python-installer | Build distributable Python installer packages |
Business & Operations (20 skills)
| Skill | Description |
|---|---|
| carrier-relationship-management | Logistics carrier relationship & SLA management |
| customer-billing-ops | Billing automation & subscription ops |
| customs-trade-compliance | Import/export compliance & HS code classification |
| email-ops | Email automation, filtering & CRM integration |
| energy-procurement | Energy market procurement & contract analysis |
| enterprise-agent-ops | Enterprise-scale agent deployment & governance |
| finance-billing-ops | Finance billing pipeline & reconciliation |
| google-workspace-ops | Google Workspace (Docs, Sheets, Drive) automation |
| healthcare-cdss-patterns | Clinical decision support system patterns |
| healthcare-emr-patterns | Electronic medical records integration |
| healthcare-eval-harness | Healthcare AI evaluation & safety harness |
| inventory-demand-planning | Demand forecasting & inventory optimization |
| investor-materials | Pitch decks, one-pagers & investor data rooms |
| investor-outreach | Investor outreach sequencing & CRM |
| jira-integration | Jira API automation & project management |
| knowledge-ops | Knowledge base management & retrieval ops |
| lead-intelligence | Lead enrichment & sales intelligence automation |
| logistics-exception-management | Logistics exception detection & resolution |
| messages-ops | Messaging platform automation (Slack, Teams) |
| product-capability | Product capability mapping & gap analysis |
| product-lens | Product strategy analysis framework |
| production-scheduling | Manufacturing & production schedule optimization |
| project-flow-ops | Project management workflow automation |
| quality-nonconformance | Quality NCR tracking & corrective actions |
| returns-reverse-logistics | Returns management & reverse logistics |
| strategic-compact | Strategy documents & executive briefings |
Specialty & Misc (18 skills)
| Skill | Description |
|---|---|
| ascii-game-dev | ECS architecture & terminal rendering |
| blueprint | Project blueprint & scaffolding templates |
| cv-latex-workspace | Multi-template CV with pdflatex/xelatex |
| data-science-workflow | Reproducible data science project structure |
| data-visualization | Publication-quality plots with Scipy/Seaborn |
| desktop-gui-dev | Python GUIs with CustomTkinter & PyQt6 |
| episode-structure-45min | 45-min 3-part episode format & pacing |
| financial-data-science | OpenBB, Pandas-TA & QuantStats stack |
| howto-documentation | Diátaxis framework & technical writing |
| image-enhancement | Neural upscaling, denoising & post-processing |
| imagemagick-reference | ImageMagick input normalization & operations |
| imagemagick-technical | PDF conversion, color space & FX operator |
| method-of-loci | Spatial memory & Blender loci construction |
| nutrient-document-processing | Document AI & structured data extraction |
| pdf-form-filling | Automated PDF form filling & generation |
| pdf-rendering-engines | PDF rendering stack comparison & selection |
| storytelling-clil-education | CLIL educational storytelling & Leitner SRS |
| visa-doc-translate | Official document translation for visa applications |
Connect this knowledge base to any MCP-compatible AI assistant. Two transports available — stdio (Claude Code CLI) and HTTP (everything else).
cp bin/mcp-server/com.agent-constitution.mcp.plist ~/Library/LaunchAgents/
launchctl load ~/Library/LaunchAgents/com.agent-constitution.mcp.plist
curl http://localhost:8765/health # → {"status":"ok","skills":343}| Tool | Config |
|---|---|
| Claude Code CLI | Zero config — stdio works out of the box |
| Cursor | ~/.cursor/mcp.json → "url": "http://localhost:8765/sse" |
| VS Code | Continue.dev extension → ~/.continue/config.json |
| Antigravity IDE | Continue.dev extension → same ~/.continue/config.json |
| JetBrains | Continue plugin → same ~/.continue/config.json |
| Gemini CLI | ~/.gemini/settings.json → "httpUrl": "http://localhost:8765/mcp" |
| Tool | Description |
|---|---|
list_skills |
List all 343 skill names |
get_skill |
Read any skill — e.g. get_skill("fastapi-best-practices") |
get_rules |
Get global repository rules |
run_<workflow> |
Execute a workflow |
See MCP Server README for full per-IDE setup instructions.
This repository's skill catalog includes content adapted from ECC — Everything Claude Code by affaan-m (MIT). If a skill isn't found here, ECC is a good next place to look.
Built with strict adherence to the Prompt-Driven Development methodology.