AGENT SKILL

rebase-regrade

Judge whether a rebase conflict resolution that changed the feature's own diff warrants regrading prd_audit or architecture_review_as_built.

License 许可
Apache-2.0
最近更新
2026-11-11
Stars
★ 7
Forks
2

01 前言 Front Matter

name
rebase-regrade
phase
ship
description
Judge whether a rebase conflict resolution that changed the feature's own diff warrants regrading prd_audit or architecture_review_as_built.
enforcement
gating
disable-model-invocation
true

02 资源文件 1

03 SKILL.md 正文

Judgement policy

A rebase completed, but its conflict resolution changed the feature's own contribution: the feature's patch after the rebase differs from its patch before it. The feature's documents (plan, stories, specs, cited ADRs) did not change. The engine supplies, per changed file, the feature patch before and after the rebase, plus the candidate gates whose earlier PASS would otherwise be kept. Answer one question per candidate: could this change plausibly alter that gate's grade?

  • prd_audit grades whether the feature delivers its stories' acceptance criteria. Regrade it when the resolution adds, removes, or alters behavior a criterion depends on.
  • architecture_review_as_built grades whether the as-built design conforms to the plan and cited ADRs. Regrade it when the resolution changes structure, boundaries, data flow, or a dependency a decision constrains.

Do not regrade for changes that preserve behavior and structure: re-ordered or re-indented hunks, renamed locals, import reshuffles, upstream identifiers adopted verbatim, or merged context that leaves the feature's own lines semantically intact. When the supplied delta is marked truncated and the omitted part could matter, regrade.

Do not read files, run commands, or use any transcript. Judge only the supplied delta.

Result contract

Return exactly one JSON object and no surrounding prose:

{ "regrade": true, "gates": ["prd_audit"], "rationale": "..." }

regrade is a boolean. gates names only supplied candidate gates: non-empty when regrade is true, empty when false. rationale is a non-empty sentence naming the change that decided it. Any other shape is rejected, and the engine then reopens every candidate.

04 同仓库更多 Skills

jstoup111
jstoup111/ai-conductor

build-review-dependency-hygiene

Judge whether a feature diff adds unnecessary, insufficiently pinned, or license-incompatible dependencies.

★ 7
jstoup111
jstoup111/ai-conductor

writing-system-tests

Use BEFORE implementing any feature that has stories in .docs/stories/ — generates failing acceptance specs from acceptance criteria as the RED phase of TDD. Generates HTTP/request-level acceptance tests for headless/API projects, end-to-end UI tests for projects with a frontend, using the project's own test framework and directory conventions.

内容创作者前端开发写作与文案
★ 7
jstoup111
jstoup111/ai-conductor

verify-claims

Use only when an active workflow is about to rely on a nontrivial factual claim or assumption for a spec, plan, ADR, finding, or code decision. Do not invoke for casual questions, status reports, restatements, or trivially verifiable mechanics.

★ 7
jstoup111
jstoup111/ai-conductor

tdd

Use when implementing any feature or bugfix. Five-step cycle: RED → DOMAIN → GREEN → DOMAIN → COMMIT. Enforces test-first development with domain integrity review at every phase boundary.

★ 7
jstoup111
jstoup111/ai-conductor

stories

Use only within active engineer/conduct DECIDE after architecture approval, when a committed `.docs/stories` acceptance artifact is required. Do not invoke for general user-story examples, requirements discussion, or direct implementation requests.

★ 7
jstoup111
jstoup111/ai-conductor

stack-land

Use when an operator needs to land reviewed manual changes on the correct closed child of a stacked feature. Operator-invoked only: inspect ownership, propose the safe child with evidence, and require confirmation before opening the landing.

★ 7