Git Memory Index

Index a Repository with git-memory

When to Use

• Starting git-memory on a new repository for the first time
• After a major history rewrite or large batch of commits
• User explicitly asks to (re-)index a project
• Setting up git-memory for a team member's repo

Workflow

1. Confirm the repo path with the user (default: current directory)
2. Run a dry-run first to show what will be indexed
3. Ask if the user wants to limit commit count (useful for very large repos)
4. Run the real index
5. Run git-memory install to register the MCP server

Checklist

- [ ] Confirmed repo path
- [ ] Ran dry-run to preview commit count and filter results
- [ ] Agreed on --limit if repo has >1000 commits
- [ ] Completed indexing with 0 errors
- [ ] MCP server registered in .claude.json
- [ ] Reminded user to restart Claude Code

Commands

Dry run (preview what will be indexed)

git-memory index \
  --repo-path /path/to/repo \
  --user-id my-repo-name \
  --dry-run

Full index

git-memory index \
  --repo-path /path/to/repo \
  --user-id my-repo-name
# Add --limit 500 for large repos (indexes 500 newest commits)

Install Claude Code plugin

git-memory install \
  --repo-path /path/to/repo \
  --user-id my-repo-name

Or from within the repo directory

cd /path/to/repo
git-memory index --user-id my-repo-name
git-memory install --user-id my-repo-name

Expected Output

Found 42 commits to evaluate
Done. {"total_evaluated":42,"stored":39,"skipped_irrelevant":3,"skipped_duplicate":0,"errors":0}

── Indexing Summary ─────────────────────────────
  total_evaluated           42
  stored                    39
  skipped_irrelevant        3
  skipped_duplicate         0
  errors                    0

Notes

• Skipped irrelevant: Commits with no signal keywords (e.g. "k", "wip", "merge") are skipped by design. This keeps the index signal-rich.
• Deduplication: Safe to run multiple times — ChromaDB deduplicates by commit hash.
• Large repos: For repos with >10k commits, use --limit 1000 to index the most recent 1000.
• user-id: Use a distinct ID per repo so multiple repos can coexist in the same index.
• Embed providers: Defaults to Ollama (nomic-embed-text). Set GITMEMORYEMBED_PROVIDER=openai for OpenAI or =fastembed for fully local (no Ollama needed).

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: gunesbizim
• Source: gunesbizim/agent-smith
• License: MIT

Install and usage instructions live in the source repository linked above.