← ClaudeAtlas

repo-map-zoom-outlisted

Map unfamiliar code areas before implementation by identifying entry points, ownership boundaries, callers, data flow, ADRs, validation, and likely edit surfaces. Use when the user asks to understand a subsystem, zoom out, map a repo area, or prepare for a risky change.
stark-ai-de/agent-skills · ★ 5 · AI & Automation · score 77
Install: claude install-skill stark-ai-de/agent-skills
# Repo Map Zoom Out ## Goal Build a compact map of an unfamiliar code area so implementation starts with the right boundaries, dependencies, and validation path. ## When to use - The user asks to understand a subsystem, route, package, workflow, or architecture boundary. - A change looks risky because callers, data flow, or ownership are unclear. - An agent needs context before editing but should avoid broad file dumping. ## When not to use - The user needs a direct code fix and the relevant files are already known. - The task is a full repo audit; use `repo-health-audit`. - The user wants an ADR; use `adr-writer` after the decision is clear. ## Inputs to inspect - Repo tree, package files, routing files, tests, ADRs, domain docs, and validation docs. - Existing code search results for entry points, callers, imports, and tests. - CodeGraph or ast-grep outputs when available, but do not require them. ## Workflow 1. Define the target area and the question the map must answer. 2. Identify entry points, exported APIs, routes, commands, jobs, or event handlers. 3. Trace likely callers, callees, imports, data models, and side effects. 4. Locate tests, fixtures, validation commands, docs, and relevant ADRs. 5. Produce a compact map with likely edit surfaces and risk points. 6. Recommend next implementation or investigation steps. ## Safety rules - Do not read broad files just to fill space; prefer targeted search and source ranges. - Do not treat naming similarity as pro