Use when looking for past work on specific FILES, understanding why code was written, or when user references "what we did before" on a file/feature. Uses file-history tracking for accurate results. PREFER THIS over episodic-memory for file-specific queries.
This skill inherits all available tools. When active, it can use any tool Claude has access to.
Use this for FILE-SPECIFIC history queries. For general semantic search across all conversations, use episodic-memory instead.
USE file-history when:
DON'T use this when:
Dispatch the history-context agent:
Task tool:
description: "Find history for [file/feature]"
prompt: "Find context for [specific file or feature]. Focus on [decisions/implementation/rationale]."
subagent_type: "history-context"
Example prompts:
The agent will return:
| Aspect | file-history | episodic-memory |
|---|---|---|
| Accuracy | Uses Claude's file-history tracking | Semantic search (fuzzy) |
| Speed | Indexed, instant | Searches full conversations |
| File mapping | Knows exactly which sessions edited which files | Guesses from content |
| Git dependency | None | None |
| Project scope | Per-project | Cross-project |
Uses ~/.claude/commands/claude-history.py which:
~/.claude/file-history/ to build file→session index~/.claude/projects/ for session metadata