Agent-Ready Analysis

Evaluate repository maturity for AI agent collaboration using the 10 Pillars / 5 Levels model.

What Makes This Different

Traditional scanners: Check if files exist (README.md ✓) Agent-Ready v0.0.2: Assess quality and AI-friendliness (README clear? AGENTS.md actionable? MCP configured?)

Key differentiator: The agent_config pillar - we evaluate Agent Native configurations that no other tool checks:

•.claude/ directory (settings, commands, hooks)
•.cursorrules / .cursor/rules
•mcp.json and MCP server implementations
•Multi-agent collaboration configs
•Autonomous workflow definitions

Analysis Framework

10 Pillars (v0.0.2)

#	Pillar	Focus
1	docs	README, AGENTS.md, API docs
2	style	Linting, formatting, types
3	build	Build scripts, CI/CD
4	test	Unit, integration, coverage
5	security	Secrets, dependabot, SAST
6	observability	Logging, tracing, metrics
7	env	.env.example, devcontainer
8	task_discovery	Issue/PR templates
9	product	Feature flags, analytics
10	agent_config	Agent Native configs

5 Levels

Level	Name	Score Range	Description
L1	Functional	0-20	Basic functionality works
L2	Documented	21-40	Essential documentation
L3	Standardized	41-60	Standard practices
L4	Optimized	61-80	Advanced automation
L5	Autonomous	81-100	Self-improving, AI-ready

How to Analyze

Step 1: Quick Baseline (Optional)

For fast file-existence checks, run the CLI:

bash

npx agent-ready scan . --output json

This gives you a quick snapshot but only checks file existence, not quality.

Step 2: Deep Analysis

Use Read/Glob/Grep tools to analyze each pillar:

•
Discover project structure
code
```
Glob: **/*.{json,yml,yaml,md,ts,js}
```
•
Read key files
- •README.md - Project overview
- •package.json - Scripts, dependencies
- •AGENTS.md - Agent instructions
- •.claude/ - Claude Code config
•
Evaluate quality using references/scoring-rubric.md
•
Follow patterns in references/analysis-patterns.md

Step 3: Generate Report

Output format:

markdown

## Agent Readiness Report

**Level: L3** (Standardized)
**Overall Score: 72/100**

### Pillar Breakdown
| Pillar | Score | Key Finding |
|--------|-------|-------------|
| docs | 85/100 | README clear, missing API docs |
| agent_config | 45/100 | AGENTS.md exists, no MCP |
| test | 65/100 | Good unit tests, no e2e |
...

### Top Recommendations
1. Configure MCP server (+15 agent_config)
2. Add integration tests (+10 test)
3. Add API documentation (+5 docs)

Agent Configuration Analysis (New in v0.0.2)

What to Look For

L1 - Basic:

•AGENTS.md or CLAUDE.md exists
•.gitignore covers .claude/, .cursor/

L2 - Structured:

•.claude/settings.json
•.claude/commands/*.md
•.cursorrules
•.aider.conf.yml
•.github/copilot-instructions.md

L3 - MCP Integration:

•mcp.json configured
•MCP server implementation
•Claude hooks defined

L4 - Advanced:

•Multi-agent collaboration
•Context injection system
•Permission boundaries

L5 - Autonomous:

•Autonomous workflows
•Self-improvement mechanisms

Quality Assessment

For AGENTS.md, check:

•Does it explain key commands?
•Does it describe architecture?
•Does it list code conventions?
•Does it specify files to ignore?
•Is it actionable for AI agents?

For .claude/settings.json, check:

•Are permissions properly restricted?
•Are dangerous commands blocked?
•Are allowed tools specified?

CLI Reference (For Quick Scans)

bash

# Basic scan
npx agent-ready scan .

# JSON output
npx agent-ready scan . --output json

# Generate missing files
npx agent-ready init . --level L2

# Preview what would be created
npx agent-ready init . --level L2 --dry-run

References

•Scoring rubric: references/scoring-rubric.md
•Analysis patterns: references/analysis-patterns.md
•Pillar details: references/pillars.md
•Level requirements: references/levels.md