implementation: let implement-dispatch run a spec'd item list without a PLAN.md
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
- 32/100
- Tipo de issue
- Nueva funcionalidad
- Claridad
- Bastante claro
- Estado de actividad
- Activo
- Stack tecnológico
- markdown, shell
Línea de trabajo
Start from the existing /implementation:implement-dispatch skill and SKILL.md (item 11 on quoting PLAN.md ## Design, phases, and implementation:phase-verifier). Trace how briefs, phases, and verifiers currently require an approved PLAN.md. Design an item-list mode (e.g. --items <report path>) that groups file:line items into disjoint workers, injects a shared PR-lifecycle block, uses per-item acceptance checks instead of phase-verifier, and defines a denied-edit return shape. Related: #5764. Done when dispatch can run from a scoped audit/report without writing PLAN.md.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
Problem
/implementation:implement-dispatch assumes an approved PLAN.md: briefs quote its ## Design section (SKILL.md item 11), phases come from it, and each phase boundary runs implementation:phase-verifier. In the 2026-10-04 CI-performance session the work arrived instead as audit or report output that was already scoped to file:line with acceptance evidence:
- a version-consumer audit listing 10 required changes with file:line targets;
- a selection-gap plan (F1-F9) with file:line targets and expected gap counts;
- verifier findings to fold back into open PRs.
Writing a PLAN.md for that would only restate the report. So the session bypassed the skill and hand-dispatched general agents, each told to spawn its own fresh-context verifier and carry its own PR through ready → CI → review threads → merge. That worked, but every brief re-typed the same lifecycle rules, and three recurring gaps had no skill support:
- File-ownership fencing across parallel workers. Two workers started in overlapping areas. The coordinator had to reassign ownership by hand mid-flight and relay one worker's half-built script to another.
- PR lifecycle per worker. Each worker needed: draft, verifier, ready, an empty commit when a draft-era ci-status failure lingers, review threads resolved with resolveReviewThread as a separate call, merge-queue
--match-head-commit, version-conflict resolver, re-queue once on a known flake. All of it was re-specified in every brief. - Classifier denials in a worker. A worker cannot see the owner's in-session approval, so an approved edit was denied in the worker and had to be escalated and applied in the main session. The skill has no "escalate the exact edit to the coordinator" return shape.
Ask
Add a context-specific entry point that keeps the dispatch discipline but drops the plan-document requirement when the input is already scoped. For example implement-dispatch --items <report path> or an "item-list mode":
- Input: a list of items, each with file:line targets and an acceptance check, from an audit, review or report file.
- Group items into workers by disjoint file sets, and record ownership so a second worker is refused an owned path.
- One fresh-context verifier per resulting PR, with criteria taken from each item's acceptance check. This replaces phase-verifier, which needs a plan.
- A shared PR-lifecycle block (the steps in 2) injected into every brief, instead of hand-copied.
- A standard return shape for "denied edit, here is the exact diff" so the coordinator can apply it where the approval is visible.
Related: #5764 (dispatch independent plan phases in parallel by default).
Evidence
Session dba79e68-af19-4ecc-a0ce-69d8b1dbefbb (transcript under ~/.claude/projects/C--Users-KyleSexton/). Resulting PRs include #6343-#6346, #6359, #6369, #6370 and the phase-3 PRs for #6006.
- Lenguaje dominante
- Shell
- Estrellas
- 22
- Forks
- 2
- Merge medio
- 5 h 11 min
- PR fusionados (30 d)
- 838
Preparar el entorno
- Sin Dockerfile ni archivo de Docker Compose
- Sin plantilla de pull request
- Leer la guía de contribución
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de melodic-software/claude-code-plugins
-
good first issue needs-triage priority: medium
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
melodic-software/claude-code-plugins#6631 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
needs-triage
Dificultad 2/5 1-3 horas Aptitud para principiantes 78/100
melodic-software/claude-code-plugins#6547 ·
Los mantenedores suelen responder en 1 día
-
needs-triage
Dificultad 2/5 1-3 horas Aptitud para principiantes 76/100
melodic-software/claude-code-plugins#6535 ·
Los mantenedores suelen responder en 1 día
-
test_comment_census.py: SccArgv flag-shaped-filename test errors on Windows (#!/bin/sh scc shim)Abiertogood first issue needs-triage priority: low
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
melodic-software/claude-code-plugins#6532 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
good first issue needs-triage priority: low
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
melodic-software/claude-code-plugins#6390 · 1 comentario ·
Los mantenedores suelen responder en 1 día
Todos los issues de melodic-software/claude-code-plugins
Issues similares
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
gnosis/gnosis_vpn#540 ·
Los mantenedores suelen responder en 1 día
-
Lid close does not lock the session on Apple Silicon (lid-close bind skips omarchy-system-lid-close)Abierto
Dificultad 1/5 1-3 horas Aptitud para principiantes 90/100
omacom/omarchy-mac#701 · 1 comentario ·
Los mantenedores suelen responder en 1 día
-
A 20.x release after 21.0.0 would move `latest` back to 20.x, and `next` stays on the release candidatePosiblemente ocupada @armando-navarro la tomó hoy. Abiertocomp: build/pipeline type: bug version: current (v17+)
Dificultad 2/5 1-3 horas Aptitud para principiantes 85/100
angular/angularfire#3790 ·
Los mantenedores suelen responder en 3 días
-
ready-for-agent
Dificultad 2/5 1-3 horas Aptitud para principiantes 85/100
LucasSantana-Dev/Lucky#2698 ·
Los mantenedores suelen responder en 1 día
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
collabnix/awesome-mcp-lists#179 ·
Los mantenedores suelen responder en 1 día