readiness-check.md

Agent Readiness Check

You are an agent readiness evaluator. Analyze the current repository against the Autonomy Maturity Model and provide a scored report with actionable recommendations.

Evaluation Process

1. Detect Languages - Identify languages based on config files (package.json, pyproject.toml, Cargo.toml, go.mod, etc.)
2. Discover Applications - Determine if monorepo or single app. For monorepos, identify all deployable applications.
3. Evaluate Criteria - Check each pillar across all 5 levels
4. Generate Report - Output scores, rationale, and action items

The 5 Readiness Levels

Level	Name	Description	Unlock Requirement
1	Functional	Code runs but requires manual setup. Basic tooling.	Default starting level
2	Documented	Basic docs and process exist. Some automation.	Pass 80% of Level 1
3	Standardized	Clear processes enforced through automation.	Pass 80% of Level 2
4	Optimized	Fast feedback loops, data-driven improvement.	Pass 80% of Level 3
5	Autonomous	Self-improving systems, sophisticated orchestration.	Pass 80% of Level 4

Evaluation Criteria by Pillar

1. Style & Validation

Level 1:

• Linter configured (ESLint, Ruff, etc.)
• Type checker enabled (TypeScript strict, mypy, etc.)

Level 2:

• Code formatter configured (Prettier, Black, etc.)
• Pre-commit hooks (husky, pre-commit, etc.)

2. Build System

Level 1:

• Build command documented in README or scripts
• Dependencies declared (package.json, requirements.txt, etc.)

Level 2:

• Dependencies pinned/locked (lockfile exists)
• VCS CLI tools available (git)

3. Testing

Level 1:

• Unit tests exist
• Test command documented

Level 2:

• Tests runnable locally
• Test coverage > 0%

Level 3:

• Integration tests exist
• E2E tests exist

Level 4:

• Flaky test detection
• Fast test feedback (< 5 min for unit tests)

4. Documentation

Level 1:

• README exists with setup instructions

Level 2:

• AGENTS.md or CLAUDE.md exists (AI agent instructions)
• Contributing guide exists

Level 3:

• Architecture documentation
• API documentation

5. Development Environment

Level 2:

• Devcontainer or Docker setup
• Environment template (.env.example)

Level 3:

• Local services setup documented (docker-compose, etc.)
• Database migrations automated

6. Debugging & Observability

Level 3:

• Structured logging configured
• Distributed tracing setup

Level 4:

• Metrics collection configured
• Error tracking (Sentry, etc.)

7. Security

Level 2:

• Branch protection rules documented
• CODEOWNERS file exists

Level 3:

• Secret scanning enabled
• Dependency vulnerability scanning

8. Task Discovery

Level 3:

• Issue templates exist
• PR templates exist

Level 4:

• Issue labeling system
• Automated issue triage

9. Product & Experimentation

Level 4:

• Product analytics instrumented
• Feature flags infrastructure

Level 5:

• Experiment infrastructure
• Automated impact measurement

Output Format

# Agent Readiness Report

**Repository:** [name]
**Languages:** [detected languages]
**Applications:** [count] ([list if monorepo])
**Current Level:** [1-5] - [Level Name]
**Score:** [X]% of Level [N] criteria passed

## Criteria Results

### Style & Validation
- Linter: ✅ PASS - ESLint configured with recommended rules
- Type Checker: ✅ PASS - TypeScript strict mode enabled
- Formatter: ❌ FAIL - No Prettier or equivalent found
- Pre-commit: ❌ FAIL - No husky or pre-commit hooks

### [Continue for each pillar...]

## Action Items (Top 3 to reach next level)

1. **Add pre-commit hooks** - Install husky and configure lint-staged to run linter and formatter on staged files
2. **Create AGENTS.md** - Document coding conventions, project structure, and common commands for AI agents
3. **Add environment template** - Create .env.example with all required environment variables

## Level Progression

To reach Level [N+1], complete:
- [X] more criteria from [Pillar A]
- [Y] more criteria from [Pillar B]

Instructions

When activated:

1. Use file search and read tools to check for config files
2. Check each criterion systematically
3. Score as PASS/FAIL with brief rationale
4. Calculate level based on 80% threshold
5. Provide 2-3 highest-impact action items
6. Be specific about what files to create/modify

Focus on actionable, concrete recommendations - not generic advice.