FAQ

Short answers about what Stacklit reads, when it helps, how accurate its dependency edges are and how it fits with git and assistants.

Does it read my code? Yes, locally, to extract structure with tree-sitter. Nothing leaves your machine unless you pass --summary (see Privacy).

Does it help on small repos? Less. If the whole repo fits comfortably in a prompt, an agent can just read it. The index pays off once exploring costs more than reading a map, and it saves the most when the agent would otherwise read several large files to confirm a location.

When is grep better? When you already know the exact identifier. A single grep is cheaper than loading a map. The index helps with questions grep cannot answer in one step, such as “what depends on this module” or “where would a new adapter go”.

How accurate are the dependency edges? Module-level edges are built from import statements and are good enough to orient an agent, but not exact. In the benchmark, 4 of 7 “what depends on X” lookups were fully correct. Known weak spots: test files counted as dependents, relative imports resolved to the wrong module, and short module names that can be ambiguous. Treat the graph as a guide and confirm with a read.

What if my language is not supported? Files are still counted and language-detected, and the module map, git activity and hints still work. Exports and dependency edges need a parser.

Does the git hook slow commits down? It runs stacklit generate --quiet after each commit. On Go repos of this size that is tens of milliseconds; on Python and TypeScript repos in the benchmark it took a few seconds. If that is too slow, run stacklit generate manually or in CI instead.

Can I keep the Mermaid file out of the repo root? Yes. Set "output": {"mermaid": "docs/DEPENDENCIES.md"} in .stacklitrc.json. The git hook stages the configured path.

Can I use it with Copilot? Yes. Commit stacklit.json and tell Copilot to read it in .github/copilot-instructions.md, or paste the output of stacklit derive there.

How is it different from dumping the whole repo into a prompt? A dump is the full source. Stacklit is structure only: it is orders of magnitude smaller but cannot answer questions that need code bodies. See COMPARISON.md for a factual feature comparison.

Generated from docs/FAQ.md in the repository, the single source of truth.