identification

The join, made inspectable. Vinv's core move is connecting your code to your run — identification is that join as a CLI: list a repo's entry points, build the call tree behind each one, and overlay a recorded run to see what actually executed.

Fully deterministic — no server, no agent, no LLM. It reads the Rust code index from <repo>/.vinv/index and, where a trace exists, a recorded run.

Command	Purpose
identification consolidate <repo>	List every entry point the code defines — HTTP routes, CLI commands, background and scheduled tasks — with its handler.
identification calltree <repo> --api-id ID	Build the call tree behind one entry point (or --symbol module:qualname for an undeclared function).
identification tracemap <repo> --api-id ID	Overlay a recorded run on that tree: what executed, how often, how long — and what never ran. Takes --symbol too.
identification tracesummary <repo>	Rank every endpoint by how hard the recorded run exercised it.

// 01 · install

cd identification
uv sync
uv run identification --help

Prerequisites: the repo has been indexed (index index <repo> --store-dir <repo>/.vinv/index), and for tracemap / tracesummary a run has been recorded (via bringup or tracelens run).

// 02 · use

# 1 · list entry points
uv run identification consolidate /path/to/repo --service my-service

# 2 · the call tree behind one of them
uv run identification calltree /path/to/repo --api-id POST_checkout

# 3 · overlay the recorded run
uv run identification tracemap /path/to/repo --api-id POST_checkout

# 4 · which endpoints did the run actually hit, busiest first
uv run identification tracesummary /path/to/repo --service my-service

Results are written under <repo>/.vinv/identification/ and printed as JSON (add --json for machine-readable trees).

---

part of vinv · vinv.ai