travsr init

Install the git hook and run the first full index of the current repository.

Synopsis

travsr init [--semantic] [--force] [--jobs <N>] [--quiet] [--json]
           [--no-connect] [--allow-unsandboxed-lsif]

Flags

Flag	Default	Description
--semantic	false	Run semantic Phase B (call edges) synchronously before returning. By default Phase B runs in the background via the daemon. Use in CI or scripts that query call edges immediately after init.
--jobs <N>	CPU cores	Number of parallel parse workers
--quiet, -q	false	Suppress progress output and post-index tips
--json	false	Emit machine-readable JSON (summary on stdout, progress on stderr)
--force, --rebuild	false	Force a full rebuild, bypassing the incremental "already up to date" skip. Re-parses every file even when nothing changed on disk. Use it after changing a flag that affects semantic output, which per-file change detection does not pick up on its own
--no-connect	false	Skip auto-detecting AI coding tools and wiring them to Travsr. See travsr connect
--allow-unsandboxed-lsif	false	Let rust-analyzer run without OS sandboxing (bubblewrap on Linux, sandbox-exec on macOS) on hosts that have none. Without it, Rust falls back to structural analysis only on such hosts. Set it only for repositories you fully trust, since Phase B executes the repo's own build configuration

What it does

1. Creates .travsr/ in the repository root
2. Installs .git/hooks/post-commit (or appends to an existing hook)
3. Runs a full Tree-sitter + LSIF index of all files
4. Writes the graph to .travsr/graph.db
5. Detects the AI coding tools installed on this machine and wires each one to the repository's MCP server, unless you pass --no-connect. This is the same work travsr connect does on its own

Example

$ cd my-project
$ travsr init

✓ Git hook installed (.git/hooks/post-commit)
✓ Detecting languages...  TypeScript, Rust
✓ Indexing 1 247 files...
  ├─ Tree-sitter pass:  1.2 s
  └─ LSIF (tsc):        8.4 s
✓ Built graph: 18 432 nodes, 94 751 edges
✓ Saved to .travsr/graph.db (4.2 MB)

Run `travsr daemon start` to keep the graph fresh.

Notes

• Running travsr init again is safe: it re-installs the hook and re-indexes the repository over the existing .travsr/ directory.
• The git hook is a thin shell wrapper; it calls travsr hook-run --from-hook and exits immediately. It does not block the commit.
• Add .travsr/graph.db to .gitignore if you don't want to commit the graph to the repository.