techdocs

Plugin techdocs from marketplace greenblacked-ai in greenblacked/AI. Commands read from the repo's manifest — 2 days ago.

Skills (12)

• accessibility-conformance-report

  Turn accessibility audit results into an Accessibility Conformance Report (a completed VPAT) and a public accessibility statement that buyers and disabled users can rely on. Fixes the product version, platforms, standard edition and test method first; rates every criterion Supports, Partially Suppor

• cli-reference

  Document a command-line tool so --help, the man page, the web reference and shell completions all come from one source, the parser definition. Lays help out as a usage line, a one-line purpose, flags grouped by task with defaults and environment variables, three real examples and exit codes; states

• code-sample-testing

  Make every code sample in the documentation executable. Samples live as real source files and the page includes them by region marker, never by pasting; CI compiles and runs each one against the released package and against main, asserts its output rather than its exit code, and fails on a pasted ru

• docs-drift-detection

  Find which documentation pages a code change makes false, before it merges. Build a reviewed map, kept in the repository, from code paths, config keys, CLI flags, environment variables and endpoints to the pages that name them; have CI list the affected pages on every pull request touching mapped co

• docs-search-vocabulary

  Fix documentation findability by mining the docs search log for the gap between the reader's words and the docs' words. Export zero-result queries and sessions reformulated twice without a click, cluster them by intent, and decide each cluster: add a search synonym, rename the heading to the reader'

• docs-url-permanence

  Move, rename or restructure a documentation site without breaking inbound links. Before anything moves, harvest the URLs that matter from search console exports, access logs, links inside the product and READMEs; write the old-to-new redirect map first and test it as a table; give each page one perm

• error-catalogue

  Give every error a product shows a stable code and its own documentation page, and keep the two in step. The code is a namespaced, immutable contract that is never reused; each page states the cause, how to confirm it, the fix and whether retrying helps; the runtime message links to its page; CI fai

• openapi-diff-gate

  Catch breaking API changes from the OpenAPI document itself before they merge. Diff the last released spec against the pull request's with a structural differ such as oasdiff, never a text diff; classify each change by direction, because request and response compatibility invert — narrowing what a r

• screenshot-rot

  Stop documentation screenshots and diagrams going stale. Inventory every image in the docs and decide per image whether text, a code block or a table would serve instead; replace the hand-taken screenshots that remain with scripted captures (browser automation, seeded demo data, fixed viewport, cloc

• terminology-lock

  Make one word mean one thing across the product UI, docs, API and support. Inventory the competing terms and where each appears, choose the canonical one, and keep a glossary in the repository giving each term's definition, its banned synonyms and the reason for each ban; generate Vale substitution

• tutorial-clean-room

  Prove a tutorial, quickstart or getting-started guide works for a stranger before it ships: run it in a brand-new container or VM with no cached credentials, dotfiles or warm caches, stopwatch running, executing only what the page literally says; log every stumble as a defect with the minutes it cos

• work-to-note

  Turn finished work (a debugging session, decision, migration, review finding, tool evaluation or incident fix) into one durable note in a Markdown vault of wikilinks and frontmatter, written to pass the vault's checks first time. Picks the note type or declines when nothing is reusable, follows the

Subagents (1)

• docs-search-log-reader

  Read a documentation site-search export of many queries (CSV, JSON lines or an analytics download, tens of thousands of rows) and return the zero-result clusters and the reformulated-query clusters, each with its count, a representative query and the page clicked at the end of a chain, so the export

Similar plugins