Hacktoberfest 2026: los issues que los mantenedores marcaron para octubre, abiertos y aptos para principiantes. Explorar issues de Hacktoberfest

feat: adopt the shared docs lookup across fetching skills and agents

Cerrado
  • ¿Algo no está bien?
#6,484 1 comentario 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 1 día

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
4/5
Tiempo estimado
3-5 días
Aptitud para principiantes
64/100
Tipo de issue
Refactorización
Claridad
Bien especificado
Estado de actividad
Activo
Stack tecnológico
bash, shell
Área
documentation, tooling

Línea de trabajo

Start by reading the shared lookup delivered in #6348: lib/fetch-docs.sh, lib/docs-cache.sh, and the /discovery:read-docs skill, plus scripts/sync-shared-copies.sh. Build the inventory.txt from the caller paths listed in the issue, then update each to invoke the lookup (or add the one-line WebFetch pointer for discovery-only callers), syncing the scripts into discipline, knowledge, plugin-quality, instruction-placement and harness-memory. Done means the grep -L command over inventory.txt prints 0 and scripts/sync-shared-copies.sh --check, scripts/check-changelog-parity.sh --check, and the tests for changelog-status and check-doc-citations all exit 0.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

needs-triage priority: medium

No related issue: follow-up to #6020 (Phase 10 of the docs-cache plan); part of #6020.

Summary

#6020 asks to replace copied upstream docs content with pointers plus a cached lookup layer. #6348 (merged) delivered the shared docs cache, the fetcher (lib/fetch-docs.sh, lib/docs-cache.sh) and /discovery:read-docs. No existing skill or agent uses it yet. This issue adopts the lookup in every skill and agent that fetches upstream documentation.

Fix

Every caller below goes through the lookup (/discovery:read-docs, or lib/fetch-docs.sh and lib/docs-cache.sh from scripts). A caller that only discovers pages (finds candidate URLs, never relies on page content) keeps WebFetch and gains a one-line pointer to the rungs in the upstream-drift convention.

Each plugin that newly carries the lookup gets the synced scripts and the procedure text, plus a version bump and a CHANGELOG line (a changelog fragment for plugins listed in scripts/fragment-plugins.txt).

Freshness per caller:

  • --max-age 0 (fresh bytes required): /harness-config:audit (audit-engine.sh, check-doc-citations.sh, check-effort-pins.sh; it runs after a Claude Code update), /harness-ops:changelog (changelog-status.sh; its purpose is the newest release), and /knowledge:docpage-digest absence-establishing reads (the Anthropic docs profile binds absence to a fresh raw fetch with a recorded length). A failed forced fetch returns unread, never stale bytes.
  • Default TTL with the age shown: all other callers. The read reports retrieved (bytes first fetched) and validated (last confirmed current); age counts from validated.

Callers named in the plan:

  • discovery: research (context/discipline.md lines about ad-hoc curl and WebFetch, SKILL.md), research-deep (SKILL.md), researcher agent
  • discipline: recheck-against-upstream, recheck-against-upstream-deep, do-your-research, sweep-all (reference/batched-pass.md)
  • harness-config: audit (SKILL.md, scripts/audit-engine.sh, scripts/check-doc-citations.sh; --cache --max-age 0)
  • harness-ops: changelog (scripts/changelog-status.sh, context/read-actions.md), inventory, audit-native-overlap, known-issues (context/action-status.md)
  • plugin-quality: auditor agent
  • instruction-placement: migrate (SKILL.md, reference/sources.md)
  • harness-memory: audit (context/update.md)
  • knowledge: docpage-digest (context/anthropic-docs-profile.md), map-corpus (discovery/link-map-format.md)

Plugins that need the synced scripts added: discipline, knowledge, plugin-quality, instruction-placement, harness-memory.

Verification

  • A written inventory file lists every path above, one per line. grep -L -F -e '<exact pointer line>' -e 'docs-cache.sh' -e 'fetch-docs.sh' $(cat inventory.txt) | wc -l prints 0, where the pointer line is the fixed one-line WebFetch pointer and the other two patterns match a lookup invocation. A bare /discovery: match does not count, since every discovery skill already contains it.
  • scripts/sync-shared-copies.sh --check exits 0.
  • scripts/check-changelog-parity.sh --check exits 0.
  • Per-caller tests for the --max-age 0 scripts (changelog-status, check-doc-citations) still pass.

Related

  • #6020 (umbrella)
  • #6348 (shared docs cache, fetcher, /discovery:read-docs)
  • Sibling follow-ups: the copied-content sweep (Phase 11) is filed separately.
Lenguaje dominante
Shell
Estrellas
22
Forks
2
Merge medio
5 h 11 min
PR fusionados (30 d)
838

Preparar el entorno

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de melodic-software/claude-code-plugins

Todos los issues de melodic-software/claude-code-plugins

Issues similares

Más issues de Shell/Bash

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.