docsynclisted

# Docsync Compare documentation against the code it describes. Find where the docs say one thing and the code says another. Propose fixes grounded in current code, not in what the docs *should* have said. ## Repo context `/techne:docsync` audits a doc that may live in a **different repo than your CWD** (e.g. `/techne:docsync ../velocity-fl/README.md`). Skill-context must therefore come from the *target's* repo, not CWD. A load-time `` !`cat .claude/skill-context.md` `` injection can't do this — it runs before the target path is parsed, so it always reads CWD. Resolve the repo root from the doc-path argument instead: ```bash git -C "$(dirname "<target-doc>")" rev-parse --show-toplevel # target repo root; with no path arg this resolves to the CWD repo ``` Then `Read` `<root>/.claude/skill-context.md`. Its `## repo` section names the CLI entrypoint and runner module the repo exposes (so command and `make`-target claims verify against the right code); `## slop_ground_truth` names where quantitative claims must trace. No `.claude/skill-context.md` at the target root → fall back to generic verification defaults. ## Checkable claims - **Commands** — `uv sync`, `make lint`, `<cli-entrypoint> <cmd>` (from injected `repo.cli_entrypoint`). Verify the target exists in the Makefile / runner module / workflow. - **File paths** — any path referenced in prose. Verify it exists. - **Function / class signatures** — `SomeClass(config: Config, ...)`. Verify against source. - **Config ke