Skip to content

Latest commit

 

History

History
33 lines (23 loc) · 2.41 KB

File metadata and controls

33 lines (23 loc) · 2.41 KB

Scripts

Repository automation: the task-doc tooling, the one-time clone setup, and the CI helpers. None of it is part of the Go module — these run by hand, from .githooks/, or from .github/workflows/ci.yml.

Task docs

Script Invoked as Role
tasks-archive.sh bash scripts/tasks-archive.sh Moves finished todos from docs/tasks/active/ into docs/tasks/archive/YYYY/MM/, bucketed by each todo's **Created** line. A todo has to clear two bars: no unchecked boxes, and a parseable **Created** date — one missing the date is warned about and left alone. A matching -lessons.md rides along if it exists; a todo without one still moves. Neither bar reads the prose, so check a todo's Review section before trusting the result.
tasks-index.sh bash scripts/tasks-index.sh Regenerates docs/tasks/README.md and docs/tasks/archive/README.md. Never hand-edit those two. docs/tasks/active/README.md is hand-written prose and is left alone.

Both take an optional tasks directory argument, defaulting to docs/tasks.

Verification

Script Invoked as Role
verify-doc-links.mjs node scripts/verify-doc-links.mjs Walks the documentation graph from CLAUDE.md, AGENTS.md, and README.md, and fails on a link that resolves to nothing. Archived task records are reached but not walked — a finished task's citations are a record of what was true then. Run by the Docs workflow, which exists separately from ci.yml because that one ignores **/*.md.

Setup

Script Invoked as Role
setup.sh bash scripts/setup.sh Points core.hooksPath at .githooks/. Run once per clone.

Directories

Directory Contents
test/ node --test suites for the scripts above, run by the Docs workflow. Invoke with the glob — node --test 'scripts/test/**/*.test.mjs' — since passing the directory makes Node try to load it as a module. Every case plants its tree under the OS temp directory and shells out to nothing.
ci/ Helpers for the benchmark and load-test jobs in .github/workflows/ci.yml. parse-bench.js diffs Go benchmark output against the base run, parse-load.js does the same for k6 output, and both render a markdown comparison table. post-comment.sh posts that table on the PR, updating its previous comment in place via an HTML marker instead of stacking a new one per run.