一键在 Manus 中运行任何 Skill

fix-runner

星标2

分支0

更新时间2026年7月6日 21:07

Use after evaluator flags failures to run .specify/gates/run-gates.sh and apply minimal fixes, looping until gates are green. Does not invent new gates.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

CaioWF

CaioWF/keel

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

fix-runner

When: evaluator reported a FAIL, or gates are suspected red, after implement-feature has touched code for the active feature.

Template: none — this skill edits existing source/test files to make gates pass, it does not produce a spec-style artifact.

Output: minimal code/test edits; a final gate run that is green (or an explicit report of what remains red and why).

Steps:

Run .specify/gates/run-gates.sh from the repo root.
If it reports "no checks detected", there is nothing for this skill to fix at the gate level — fall back to whatever evaluator flagged directly (e.g. an unmet acceptance criterion) and fix that instead.
For each failing gate, read its output and apply the MINIMAL fix that addresses the root cause — do not refactor unrelated code, do not add new lint rules, scripts, or gates beyond what run-gates.sh already runs (it picks up npm run lint/test/build or make lint/test automatically; do not hand-roll alternatives).
Re-run .specify/gates/run-gates.sh after each fix attempt.
Loop steps 3-4 until the script exits 0 ("all gates passed"), or until a fix attempt would require scope beyond the current task (e.g. a missing approved spec/plan) — in that case stop and report the blocker instead of working around it.
Do not touch spec.md, plan.md, or tasks.md frontmatter/approval state from this skill.

Model routing: when a gate fix is handed to a subagent (via dispatching-parallel-agents) instead of run inline, size the model to the fix's difficulty — do NOT inherit the session model by default. A single-file mechanical gate fix (lint, formatting, a missing import) is cheapest tier; a fix that needs cross-file reasoning is standard tier. See the routing-minimum-capable-model skill.

Next: evaluator, to re-confirm acceptance criteria now that gates are green; loop continues via implement-and-evaluate.

同仓库更多 Skills

同仓库

okf-enrich

CaioWF/keel

Ingest a raw source into the project's OKF knowledge bundle — distill it into concept pages, refresh the index, and log the change. keel's compounding "LLM wiki" loop.

2026-07-112

finishing-a-development-branch

CaioWF/keel

Use when implementation is complete, gates pass, and you need to decide how to integrate the work — presents structured options (merge, PR, keep, discard) and handles the chosen workflow, including worktree cleanup. Closes the SDD cycle after review-and-simplify.

2026-07-102

subagent-driven-development

CaioWF/keel

Use when executing implementation plans with independent tasks in the current session

2026-07-102

implement-and-evaluate

CaioWF/keel

Orchestrates the implement-feature, evaluator, fix-runner loop for the active feature until all Criterios de Aceitacao are met and quality gates are green.

2026-07-102

tasks-writer

CaioWF/keel

Use after the plan is approved to break the active feature's plan into a concrete task checklist, written into specs/<feature>/tasks.md from the tasks template.

2026-07-102

observability-and-instrumentation

CaioWF/keel

Use while implementing a feature that has retries, queues, background jobs, external calls, or any critical path an on-call engineer will need to reason about in production. Invoke when adding a new endpoint, dependency, or async flow, and before considering such a feature done — instrumentation is developed alongside the code, like tests, not bolted on after.

2026-07-082

fix-runner

When: evaluator reported a FAIL, or gates are suspected red, after implement-feature has touched code for the active feature.

Template: none — this skill edits existing source/test files to make gates pass, it does not produce a spec-style artifact.

Output: minimal code/test edits; a final gate run that is green (or an explicit report of what remains red and why).

Steps:

Run .specify/gates/run-gates.sh from the repo root.

If it reports "no checks detected", there is nothing for this skill to fix at the gate level — fall back to whatever evaluator flagged directly (e.g. an unmet acceptance criterion) and fix that instead.

For each failing gate, read its output and apply the MINIMAL fix that addresses the root cause — do not refactor unrelated code, do not add new lint rules, scripts, or gates beyond what run-gates.sh already runs (it picks up npm run lint/test/build or make lint/test automatically; do not hand-roll alternatives).

Re-run .specify/gates/run-gates.sh after each fix attempt.

Loop steps 3-4 until the script exits 0 ("all gates passed"), or until a fix attempt would require scope beyond the current task (e.g. a missing approved spec/plan) — in that case stop and report the blocker instead of working around it.

Do not touch spec.md, plan.md, or tasks.md frontmatter/approval state from this skill.

Next: evaluator, to re-confirm acceptance criteria now that gates are green; loop continues via implement-and-evaluate.

name	fix-runner
description	Use after evaluator flags failures to run .specify/gates/run-gates.sh and apply minimal fixes, looping until gates are green. Does not invent new gates.