dead-code-finder¶
Classify dead-code candidates in a repo — unused exports, unreachable functions, orphan files, unused dependencies — by confidence, screening out known false positives (dynamic dispatch, reflection, framework auto-wiring, re-exports, public API surface). It reviews tool output or an export/import cross-reference (no install, no delete) and reports candidates with a confidence level.
At a glance¶
| Model | inherits the session's |
| Tools | Read Grep Glob Bash |
| Author | navjyotnishant |
| Source | agents/dead-code-finder.md |
Returns content — does not write files
Like all but two of the agents, this one returns its output to the skill that spawned it, and the skill writes the file. It holds no write tools, so it cannot modify the repo even if asked to.
When it runs¶
The dead-code-finder skill detected the repo has no dedicated tool and produced an export/import cross-reference.
what code here is unused?
The skill spawns this agent with the cross-reference; it classifies candidates by confidence and filters false positives, never deleting anything.
What it returns¶
- A candidates list:
file:line/symbol · category · confidence · reason · (for excluded items) why it's not dead. - A one-line summary: N high / M medium / K low candidates, over
<scope>, via<tool | manual fallback>. - Never a delete instruction — the human decides what to remove. If you can suggest a safe verification (e.g. "grep for the string form before removing"), do.
Spawned by¶
Derived, not declared
No agent file records which skills call it — this list is recovered from the skill definitions at build time, so it cannot go stale.