Aegis — AI Development Quality Guardian

"Move fast, but don't break the blueprint."

Aegis (Greek: the divine shield of Zeus and Athena) is an AgentSkill for OpenClaw that brings structured quality assurance to AI-assisted software development.

AI writes code at the speed of thought. Aegis makes sure the thoughts are correct.

AI-assisted development is fast — dangerously fast. As projects grow, things quietly fall apart:

• Design gaps accumulate silently — each AI agent sees only its slice, nobody holds the full picture
• Mock testing creates illusions — frontend and backend mock each other, tests pass, integration explodes
• Context drifts across agents — parallel agents develop divergent understandings of the same interface
• The code isn't yours — debugging AI-generated code feels like inheriting a stranger's project
• Constraints are too loose — CLAUDE.md alone isn't enough to prevent architectural drift

┌─────────────────────────────────────────────────────────┐
│ Layer 5: Project Management │
│ PM Tool Integration — Tracking + Gap Mgmt │
├─────────────────────────────────────────────────────────┤
│ Layer 4: Verification │
│ Contract Test → Integration Test → E2E Test │
├─────────────────────────────────────────────────────────┤
│ Layer 3: Implementation │
│ CLAUDE.md Constraints + Dispatch Protocol + Review │
├─────────────────────────────────────────────────────────┤
│ Layer 2: Contract │
│ API Spec + Shared Types + Event Schema │
├─────────────────────────────────────────────────────────┤
│ Layer 1: Design │
│ Design Brief + Architecture Doc │
└─────────────────────────────────────────────────────────┘

Each layer addresses a specific failure mode. Together, they form a complete quality shield.

# One-liner: clone the cc-skill folder into your CC skills directory
git clone https://github.com/skill-forge-ai/aegis.git /tmp/aegis-clone \
 && cp -r /tmp/aegis-clone/cc-skill ~/.claude/skills/aegis \
 && rm -rf /tmp/aegis-clone
# Verify installation
ls ~/.claude/skills/aegis/SKILL.md

Or manually:

1. Download/clone this repo
2. Copy cc-skill/ → ~/.claude/skills/aegis/
3. Done — Claude Code will auto-detect the skill

Then initialize Aegis in your project:

bash ~/.claude/skills/aegis/scripts/init-project.sh /path/to/your/project

# Clone into your OpenClaw skills directory
git clone https://github.com/skill-forge-ai/aegis.git ~/.openclaw/workspace/skills/aegis

Any agent that supports the AgentSkill spec can use Aegis:

1. Clone the repo or copy the cc-skill/ folder to your agent's skill path
2. Point your agent at the SKILL.md file
3. Run init-project.sh on your project

# Run the init script in your project root
bash ~/.openclaw/workspace/skills/aegis/scripts/init-project.sh /path/to/your/project

This creates:

your-project/
├── contracts/
│ ├── api-spec.yaml # OpenAPI 3.1 spec
│ ├── shared-types.ts # Shared type definitions
│ └── errors.yaml # Error code definitions
├── docs/
│ └── designs/ # Design Briefs go here
└── CLAUDE.md # Enhanced with Aegis constraints

Follow the workflow:

Design Brief → Contract → Implementation → Verification → Delivery

See SKILL.md for the complete workflow reference.

• templates/design-brief.md — Design Brief template
• templates/claude-md.md — Enhanced CLAUDE.md template
• templates/api-spec-starter.yaml — OpenAPI starter
• templates/shared-types-starter.ts — Shared types starter

Aegis acts as your quality co-pilot. Write a Design Brief before coding, define contracts for your APIs, and let contract tests catch drift before it becomes a bug.

Aegis becomes the coordination layer. All agents share the same contract. No agent can modify the contract unilaterally — changes go through a review process, ensuring everyone stays aligned.

 ┌──────────────┐
 │ Lead Agent │
 │ (holds contract)│
 └──────┬───────┘
 │
 ┌────────────┼────────────┐
 │ │ │
 ┌─────▼─────┐ ┌───▼───┐ ┌─────▼─────┐
 │ Frontend │ │Contract│ │ Backend │
 │ Agent │ │ Repo │ │ Agent │
 └───────────┘ └───────┘ └───────────┘

Aegis works with your existing tools:

1. Contract-first — Define the interface before writing the implementation
2. Mock only at the bottom — Unit tests can mock; everything above uses real services
3. Gaps are first-class citizens — Track them, triage them, resolve them
4. Design Briefs are for humans — AI writes code fast; humans need to understand what's being built
5. Progressive strictness — Lite mode for small stuff, Full mode for critical paths

Contributions welcome! Please read CONTRIBUTING.md before submitting a PR.

MIT

Development under Aegis — AI helps you build, Aegis makes sure it stands.