| name | entrypoint-finder |
| description | Find application entry points and system startup locations using CodeGraph. |
Use this skill when you need to locate where an application or service starts execution (e.g. for debugging, instrumentation or understanding control flow).
Goal: Identify the files, functions or scripts that act as entry points into the system, whether they are web servers, background workers, CLI commands or scheduled jobs.
Shared tool policy:
- Follow
../../reference/codegraph-tool-policy.md for common CodeGraph tool selection, fallback and unavailable handling.
Task-specific tools:
- Prefer
codegraph_files, codegraph_search, codegraph_node, codegraph_callers and codegraph_explore for unclear framework wiring.
Workflow:
- Ensure the graph index is available with
codegraph_status.
- Search for common entrypoint patterns using
codegraph_search:
- Functions named
main, run, start, bootstrap.
- Files like
server.js, app.py, index.ts, cli.rb.
- Framework-specific entrypoints (e.g.
manage.py, bin/rails, cmd/server, src/index.ts).
- Inspect candidate files and symbols with
codegraph_node to confirm they are top-level commands or initialisation routines.
- Use
codegraph_callers on each candidate to see if anything calls them. True entrypoints usually have no callers and are only referenced from configuration or script definitions.
- If framework wiring is unclear, use one bounded
codegraph_explore query anchored to the candidate files or route symbols.
- Compile a list of confirmed entrypoints, including:
- Type (web server, CLI command, worker, scheduler)
- File path
- Function or class name
- Description of what it starts
- Present the list, grouping similar entrypoints together and noting any that might be dead code or unused.
Rules:
- Do not remove or edit entrypoints; this skill is for identification only.
- If the analysis is ambiguous, provide multiple candidates and explain why more investigation is needed.