| name | repo-surveyor |
| description | Use when Cartograph CLI or MCP is unavailable and you still need Cartograph-style repo orientation, task context, or documentation inputs from manual file survey. |
| allowed-tools | ["Bash","Read","Glob","Grep"] |
Use this skill only when Cartograph itself is unavailable or when you need to manually verify its outputs.
Manual workflow:
- Discover source, config, and entry files. Skip generated, vendored, and build output.
- Rank likely-important files using the same broad signals Cartograph uses:
- README, AGENTS, config, entry points, routers, schemas, models
- fan-in from imports
- exported API surface
- root-level wiring files
- Trace dependency hubs around the top-ranked files instead of reading everything.
- Build the smallest file set that satisfies the current task or documentation objective.
- Produce a doc-ready summary from that reduced set.
Output contract:
- Key files: highest-value files and why they matter
- Dependency hubs: files or modules with the strongest fan-in and wiring impact
- Minimal task context: the smallest file set needed for the current task
- Doc-ready summary: concise architecture/module notes that another agent can expand
Rules:
- Prefer
use-cartograph whenever the tool becomes available.
- Keep manual reads narrow. Do not read the entire repository by default.
- Match the same output shape as
use-cartograph so downstream agents can treat both paths the same way.
