| name | verify-a1 |
| description | Verify adoption criterion A1 (Agent Context Availability) in the current project. Reports fulfillment level 0–3. |
| allowed-tools | Bash Read |
Verify A1 — Agent Context Availability
Criterion Definition
| Level | Description |
|---|
| 0 | No agent context files; agents must infer everything from raw code |
| 1 | Basic README or project description exists; agents have a starting point but no conventions or navigation guidance |
| 2 | CLAUDE.md or AGENTS.md provides project-specific guidance: conventions, architecture overview, key commands |
| 3 | Comprehensive agent context including architecture docs structured for agent consumption AND MCP servers exposing project knowledge (requirements, diagrams, runbooks) |
Evidence to Gather
- Check for agent context files (
CLAUDE.md, AGENTS.md, or equivalent) at the project root and in .claude/. Read their content and assess depth — look for sections covering conventions, key commands, architecture overview, and navigation hints.
- Check for a
README.md and assess whether it gives an agent a meaningful starting point.
- Check the MCP server configuration (
.claude/settings.json, .mcp.json) for any servers that expose project knowledge such as requirements, diagrams, or runbooks.
- Look for a
docs/ directory or architecture documentation files that are structured for agent consumption.
Instructions
Gather the evidence described above and determine the fulfillment level for A1.
Scoring guide:
- Level 0: No CLAUDE.md, AGENTS.md, or equivalent exists. No README, or a README so minimal it provides no meaningful starting point.
- Level 1: A README or basic project description exists that tells an agent what the project is and how to navigate it at a high level. No project-specific conventions, key commands, or architecture guidance is present.
- Level 2: A CLAUDE.md or AGENTS.md (or equivalent) exists with substantive project-specific content: coding conventions, key commands (how to build, test, run), an architecture overview, or navigation hints. The file must go meaningfully beyond a README.
- Level 3: Level 2 is met AND architecture documentation is structured for agent consumption (e.g., machine-readable or well-organised in-repo docs) AND at least one MCP server exposes project knowledge (requirements, diagrams, runbooks) programmatically.
Report in exactly this format:
A1 — Agent Context Availability
- Level: [0 / 1 / 2 / 3]
- Rationale: [one or two sentences citing the specific evidence]