Agent skill

Nucleus Delivery Evidence Closure

by protect-my-hair in protect-my-hair/nucleus-marketplace

Collects prior Nucleus task evidence into a candidate-only delivery closure package and PR/MR draft text, blocking if any feature's design, review or verification evidence is missing.

MITAuto-check passedDevelopment

SKILL.md written in Chinese; this summary is our English description.

Install Nucleus Delivery Evidence Closure

skills CLI
$ npx skills add protect-my-hair/nucleus-marketplace --skill delivery-evidence-closure -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install protect-my-hair/nucleus-marketplace delivery-evidence-closure --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/protect-my-hair/nucleus-marketplace.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/nucleus/skills/delivery-evidence-closure .claude/skills/delivery-evidence-closure && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
delivery-evidence-closure
GitHub stars
164
Token cost
~1k tokens
SKILL.md length
317 words
Files
8 (incl. scripts, references, assets)
Skills in repo
18
Repo updated
First seen
Licence
MIT

At a glance

Collects prior Nucleus task evidence into a candidate-only delivery closure package and PR/MR draft text, blocking if any feature's design, review or verification evidence is missing.

  • Works in 9 steps: 读取 prior result:读取明确的 prior result… → 创建宿主任务包:把本 checklist 同步成当前会话的宿主… → 收集交付证据:读取 quality gate、test report、git… → …
  • Summarizing delivery evidence from prior Nucleus task runs into a reviewable package
  • SKILL.md covers 核心原则, Checklist, 特性开发中的角色 and 边界, plus 3 more sections
  • Runs Python scripts from its folder; calls python3

What it does

This Chinese-language skill compiles delivery evidence that already exists and is traceable; it cannot fabricate completion, human review, merge or release facts. A hard gate blocks it from preparing a PR or MR draft, or declaring anything reviewable, whenever a feature task package lacks evidence of design, design review, an implementation plan, plan confirmation, implementation, verification, multi-review or remediation and its recheck.

The workflow creates host todo items for a nine-step checklist: read the prior result package and workflow run context, sync the checklist into host tasks, collect quality-gate, test-report, git-diff and review evidence, verify each feature package's completeness, recompute spec compliance against each feature's spec anchor and changed files, block on any missing critical evidence, generate only the closure candidate and PR/MR draft text, wait for a subagent pre-review before requesting human delivery review, and finally write a result package.

A PR or MR draft never implies the PR or MR has actually been created, reviewed, merged or released. The skill's boundary is to read the explicit context package and prior evidence; it does not substitute for the host task system or a human reviewer.

When your agent uses it

  • Summarizing delivery evidence from prior Nucleus task runs into a reviewable package
  • Drafting PR or MR description text from completed feature evidence
  • Checking whether a feature task package has all required review and verification evidence before closure

Example prompts

  • “汇总本次特性任务包的交付证据,生成 PR draft。”
  • “Check whether the login-redesign feature package has all required closure evidence before drafting the PR.”
  • “Build the delivery closure candidate from the prior Nucleus result package.”

Requirements

  • A Nucleus workflow run with a prior result package and context file

Workflow steps

9 steps, taken from the first numbered list in SKILL.md.

  1. 读取 prior result:读取明确的 prior result package 和 .nucleus/context/.json;缺可信 result 时 ALERT_AND_BLOCK。
  2. 创建宿主任务包:把本 checklist 同步成当前会话的宿主 todo/task;未同步前不得生成 closure candidate 或 PR/MR draft。
  3. 收集交付证据:读取 quality gate、test report、git changed files、commit metadata 和相关 review evidence。
  4. 校验特性任务包完整性:确认每个 feature task package 都有设计、设计评审、实施计划、spec anchor、计划确认、实现、验证、multi-review、整改和整改复核证据。
  5. 重算 spec compliance:读取每个特性的 specAnchor 和 feature-development/spec-compliance-evidence.json 正文,按 changed files 重新计算 applicable…
  6. 阻塞缺失证据:任何关键证据缺失时 ALERT_AND_BLOCK;不得用 summary、result 或 PR/MR draft 补造事实。
  7. 生成收口候选:只生成 delivery closure candidate 和 PR/MR description draft;draft 只是待审文本。
  8. 等待交付人工评审:到达人审点时按 _shared/references/subagent-precheck-protocol.md 执行 subagentPreReview,确认材料可提交人工审查后,再按…
  9. 写入结果证据:构建 result package;delivery closure candidate、PR/MR draft、summary 和 result 不能替代宿主任务或人工 review,也不代表 PR/MR 已创建、已审查、已合入或已发布。

What it can do on your machine

Read from SKILL.md and the folder at commit 3c82085. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Nucleus Delivery Evidence Closure loads about 1k tokens when it runs, and up to ~1.3k if it reads all its reference files. Until then it costs about 27 tokens; SKILL.md has 317 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~27
When it runs · the whole SKILL.md, loaded when a task matches
~1k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~1.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from protect-my-hair/nucleus-marketplace at commit 3c82085, republished under its MIT licence (© protect-my-hair). 317 words, ~1,014 tokens.

Download SKILL.mdSave it as .claude/skills/delivery-evidence-closure/SKILL.md (or your agent's skills folder). This skill also uses 7 other files; get the full folder from GitHub.
name
delivery-evidence-closure
description
Use when 从 prior Nucleus result packages 生成 candidate-only 交付闭环证据和 PR/MR draft 文本。

Delivery Evidence Closure

前置:使用本 Skill 前,先按 using-nucleus 完成 Nucleus 入口识别(Claude Code 会话由插件 SessionStart hook 自动注入该纪律)。

当 coding agent 需要汇总交付证据、准备可供人工审查的交付收口材料时使用本 Skill。

核心原则

交付收口只能汇总已经发生且可追溯的证据;它不能补造开发完成、人工 PR/MR 审查、合入或发布事实。

<HARD-GATE>
任何特性任务包缺设计、设计评审、实施计划、计划确认、实现、验证、multi-review、整改或整改复核证据时,不得准备 PR/MR draft 或声明交付可审。

未创建宿主 todo/task 任务包,不得生成 delivery closure candidate、PR/MR draft 或 result package。

到达 delivery-closure-review 人审点时,按 _shared/references/subagent-precheck-protocol.md 执行 subagentPreReview 子代理预审;未取得“材料可提交人工审查”结论前,不得请求人工交付评审。 </HARD-GATE>

Checklist

启动本 Skill 后,必须先为以下每一项创建宿主 todo/task,并按顺序执行;Codex 使用计划 / 任务工具,Claude Code 使用 TodoWrite 或等价宿主 todo。每完成、阻塞、等待 review 或需要复核一项,都必须逐项更新状态。

  1. 读取 prior result:读取明确的 prior result package 和 .nucleus/context/<workflowRunId>.json;缺可信 result 时 ALERT_AND_BLOCK。
  2. 创建宿主任务包:把本 checklist 同步成当前会话的宿主 todo/task;未同步前不得生成 closure candidate 或 PR/MR draft。
  3. 收集交付证据:读取 quality gate、test report、git changed files、commit metadata 和相关 review evidence。
  4. 校验特性任务包完整性:确认每个 feature task package 都有设计、设计评审、实施计划、spec anchor、计划确认、实现、验证、multi-review、整改和整改复核证据。
  5. 重算 spec compliance:读取每个特性的 specAnchor 和 feature-development/spec-compliance-evidence.json 正文,按 changed files 重新计算 applicable rules、required evidence、UI design coverage 和公共组件 / 公共样式证据;缺 spec anchor、factsManifestHash stale、报告正文为空、blocking rule 证据缺失、UI frame 仍是 pending-export 或 required UI 缺截图 / 视觉证据时 ALERT_AND_BLOCK。
  6. 阻塞缺失证据:任何关键证据缺失时 ALERT_AND_BLOCK;不得用 summary、result 或 PR/MR draft 补造事实。
  7. 生成收口候选:只生成 delivery closure candidate 和 PR/MR description draft;draft 只是待审文本。
  8. 等待交付人工评审:到达人审点时按 _shared/references/subagent-precheck-protocol.md 执行 subagentPreReview,确认材料可提交人工审查后,再按 _shared/references/interaction-format.md 的确认型格式向用户或 PMS 呈现:先给结论(本次完成了什么、变更范围、测试状态),再展开候选材料、PR/MR draft 和缺口;缺 delivery-closure-review 时 STOP。
  9. 写入结果证据:构建 result package;delivery closure candidate、PR/MR draft、summary 和 result 不能替代宿主任务或人工 review,也不代表 PR/MR 已创建、已审查、已合入或已发布。

PR/MR draft 不代表 PR/MR 已创建、已审查、已合入或已发布。

特性开发中的角色

在特性开发场景下,只有当需求内每个特性任务包都完成时,本 Skill 才可准备 PR/MR draft 文本。完成的标准是:每个特性都有设计、设计评审、实施计划、spec anchor、计划确认、实现、验证、multi-review、整改和整改复核证据。

draft 必须汇总需求、特性、已评审设计、计划、实现、测试、review 发现、整改和证据。它不得声明人工 PR review、合入批准、版本排期或发布批准。

边界

  • 读取 .nucleus/context/<workflowRunId>.json 中明确的 context package。
  • 把 prior result、quality gate、test report、git changed file 和 commit metadata 作为证据读取。
  • 只在 .nucleus/runs/<workflowRunId>/ 下写 closure candidate、PR/MR draft、gate、blockers 和 result。
  • 即使证据不足、最终状态为 FAILED_BLOCKED,也要写 delivery-closure-candidate.json,除非 prior result / schema / path 本身不可信。
  • closure 证据完整到可供审查时返回 NEEDS_HUMAN_REVIEW。
  • 缺 result package、gate 阻塞、缺测试、变更无归属、写保护路径、legacy path、PMS 驱动模式下缺 session binding 或 schema 失败时返回 FAILED_BLOCKED。

本 Skill 不创建 PR/MR,不 commit、push、merge、release,也不关闭缺陷。

STOP 点

  • 任一特性任务包不完整:ALERT_AND_BLOCK。
  • 任一特性缺 spec anchor、factsManifestHash 失效、completion gate 证据或 spec compliance 正文覆盖:ALERT_AND_BLOCK。
  • 任一特性存在 required UI 设计源但缺 ready frame mapping、截图 / 视觉对照或公共组件 / 公共样式契约证据:ALERT_AND_BLOCK。
  • 任一特性缺 multi-review、整改或整改复核证据:ALERT_AND_BLOCK。
  • 在没有 delivery-closure-review.subagentPreReview 和真实人工事实时,请求记录 PR/MR 人工 review、合入批准、版本排期或发布批准:ALERT_AND_BLOCK。
  • 请求在 PR/MR 创建后继续批准、合入或发布:ALERT_AND_BLOCK。

禁止

  • 用 summary.md 代替 result package、test report 或 review evidence。
  • 把 PR/MR draft 写成已审查、已合入或已发布。
  • 在所有特性完成前收口整个需求。
  • 把 delivery closure candidate、summary 或 result 当成宿主任务、人工 review、PR/MR 创建、合入或发布事实;只有按 _shared/references/subagent-precheck-protocol.md 完成预审、确认材料可提交人工审查后,才可进入后续人工 review。

Runtime

bash
python3 skills/delivery-evidence-closure/scripts/delivery_evidence_closure.py close \
  --repo-root <target-repo> \
  --context <target-repo>/.nucleus/context/<workflowRunId>.json

runtime 只导入打包的 Skill 资产和 skills/_shared/nucleus_runtime/,不导入根目录 harness/ 或根目录 scripts/。

© protect-my-hair, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 7 other files (scripts, references, assets) in plugins/nucleus/skills/delivery-evidence-closure of protect-my-hair/nucleus-marketplace.

  • SKILL.md
  • agents/openai.yaml
  • assets/delivery-closure-candidate-template.json
  • assets/delivery-closure-candidate-template.md
  • assets/nucleus-delivery-closure-candidate.schema.json
  • assets/pr-mr-description-draft-template.md
  • references/closure-boundary.md
  • scripts/delivery_evidence_closure.py

Open the folder on GitHubat commit 3c82085

Compare with similar skills

Nucleus Delivery Evidence Closure next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Nucleus Delivery Evidence Closure compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Nucleus Delivery Evidence Closure this skillprotect-my-hair/nucleus-marketplace164—~1kAutomated safety check: PassMIT
CodFlow Change Workflowbighadj22/codflow354—~1.1kAutomated safety check: PassApache-2.0
Veomni ReviewByteDance-Seed/VeOmni2.2k—~1.9kAutomated safety check: PassApache-2.0
System1-Agents Self-ReviewThinkFlowLab/system1-agents131—~2.1kAutomated safety check: PassApache-2.0
Reviewing PRsportabletext/editor280—~940Automated safety check: PassMIT
Review PRvllm-project/vllm-omni7.1k—~3.8kAutomated safety check: PassApache-2.0

Similar skills

  • CodFlow Change Workflow

    bighadj22/codflow

    A step-by-step workflow for changing the CodFlow repository: read the AGENTS.md contract, respect package boundaries, verify before claiming done and keep PRs small.

    354 GitHub stars~1.1k tokensUpdated 5 days ago
    DevelopmentAuto-check passed
  • Veomni Review

    ByteDance-Seed/VeOmni

    Pre-PR code review gate. An agent skill from ByteDance-Seed/VeOmni.

    2.2k GitHub stars~1.9k tokensUpdated yesterday
    DevelopmentAuto-check passed
  • System1-Agents Self-Review

    ThinkFlowLab/system1-agents

    Prepares a local self-review report for a System1-Agents change before a PR is opened, checking the full diff, tests, docs, artifacts and claims against evidence.

    131 GitHub stars~2.1k tokensUpdated today
    DevelopmentAuto-check passed
  • Reviewing PRs

    portabletext/editor

    Review a PR in this repo against the house conventions before it is marked ready.

    280 GitHub stars~940 tokensUpdated yesterday
    DevelopmentAuto-check passed
  • Review PR

    vllm-project/vllm-omni

    Review pull requests and local branches for vllm-project/vllm-omni with a frozen snapshot, module-design ownership, feature-design overlays, targeted validation, and concise evidence-backed findings.

    7.1k GitHub stars~3.8k tokensUpdated today
    AI & LLM EngineeringAuto-check passed
  • Official

    Runs a loop on a GitHub pull request: fetch review state, triage comments into actions, implement them and resolve threads, repeating until nothing actionable is left.

    48k GitHub stars~2.2k tokensUpdated 2 days ago
    DevelopmentAuto-check passed

More from protect-my-hair/nucleus-marketplace

All 18 skills in this repo
  • Nucleus Context Validator

    protect-my-hair/nucleus-marketplace

    Validates the context file for a Nucleus V2 workflow run, discovers standalone context from explicit input, and writes a blocked result when required context is missing.

    164 GitHub stars~576 tokensUpdated 2 mo ago
    Auto-check passed
  • Nucleus Session Binding Report

    protect-my-hair/nucleus-marketplace

    Collects, freezes and validates a Nucleus V2 session binding report that records the native Claude Code or Codex session ID, transcript path and hash.

    164 GitHub stars~477 tokensUpdated 2 mo ago
    Auto-check passed
  • Nucleus Result Package Builder

    protect-my-hair/nucleus-marketplace

    Builds the result.json file for a Nucleus V2 workflow run from its context, gates, blockers and step status, and exits non-zero when the run is blocked.

    164 GitHub stars~414 tokensUpdated 2 mo ago
    Auto-check passed
  • Defect Intake and RCA Candidates

    protect-my-hair/nucleus-marketplace

    Checks whether a defect report has enough context to be accepted, then drafts candidate-only root-cause evidence for human review without confirming causes or changing code.

    164 GitHub stars~745 tokensUpdated 2 mo ago
    Auto-check passed
  • Nucleus Delivery Evidence Finalizer

    protect-my-hair/nucleus-marketplace

    Combines the evidence from completed Nucleus workflow steps into a final evidence report and result package, without scheduling steps, updating trackers or opening pull requests.

    164 GitHub stars~454 tokensUpdated 2 mo ago
    Auto-check passed
  • Feature Doc Design

    protect-my-hair/nucleus-marketplace

    A skill your agent uses when 特性开发需要直接在产品仓库 docs/features 下写可评审的特性文档和邻近设计文档。

    164 GitHub stars~895 tokensUpdated 2 mo ago
    Auto-check passed

Questions about Nucleus Delivery Evidence Closure

What does Nucleus Delivery Evidence Closure do?

Collects prior Nucleus task evidence into a candidate-only delivery closure package and PR/MR draft text, blocking if any feature's design, review or verification evidence is missing. This Chinese-language skill compiles delivery evidence that already exists and is traceable; it cannot fabricate completion, human review, merge or release facts. A hard gate blocks it from preparing a PR or MR draft, or declaring anything reviewable, whenever a feature task package lacks evidence of design, design review, an implementation plan, plan confirmation, implementation, verification, multi-review or remediation and its recheck.

When should I use Nucleus Delivery Evidence Closure?

Nucleus Delivery Evidence Closure fits situations like: summarizing delivery evidence from prior Nucleus task runs into a reviewable package; drafting PR or MR description text from completed feature evidence; checking whether a feature task package has all required review and verification evidence before closure.

How do I install Nucleus Delivery Evidence Closure in Claude Code?

Run `npx skills add protect-my-hair/nucleus-marketplace --skill delivery-evidence-closure -a claude-code`. Or copy the skill folder (plugins/nucleus/skills/delivery-evidence-closure in protect-my-hair/nucleus-marketplace) into .claude/skills/delivery-evidence-closure in your project. Claude Code loads it when a task matches its description.

How do I install Nucleus Delivery Evidence Closure in Codex?

Run `npx skills add protect-my-hair/nucleus-marketplace --skill delivery-evidence-closure -a codex`. Or copy the skill folder (plugins/nucleus/skills/delivery-evidence-closure in protect-my-hair/nucleus-marketplace) into .agents/skills/delivery-evidence-closure in your project. Codex loads it when a task matches its description.

Can I use Nucleus Delivery Evidence Closure in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add protect-my-hair/nucleus-marketplace --skill delivery-evidence-closure -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/delivery-evidence-closure, .gemini/skills/delivery-evidence-closure, .github/skills/delivery-evidence-closure and .opencode/skills/delivery-evidence-closure in your project.

What does Nucleus Delivery Evidence Closure need to run?

Going by SKILL.md and its folder, Nucleus Delivery Evidence Closure needs Python for the scripts in its folder and the command-line tools its instructions call (python3). Our summary lists: A Nucleus workflow run with a prior result package and context file.

Does Nucleus Delivery Evidence Closure access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Nucleus Delivery Evidence Closure safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Nucleus Delivery Evidence Closure use?

Nucleus Delivery Evidence Closure is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Nucleus Delivery Evidence Closure use?

About 1k tokens (SKILL.md is roughly 4.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 240 tokens, read only when the agent opens those files.

What are the alternatives to Nucleus Delivery Evidence Closure?

Skills that share tags, products or a category with Nucleus Delivery Evidence Closure: CodFlow Change Workflow (bighadj22/codflow, 354 stars), Veomni Review (ByteDance-Seed/VeOmni, 2.2k stars), System1-Agents Self-Review (ThinkFlowLab/system1-agents, 131 stars) and Reviewing PRs (portabletext/editor, 280 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Nucleus Delivery Evidence Closure?

protect-my-hair (a GitHub user) maintains it in protect-my-hair/nucleus-marketplace, which has 164 GitHub stars. The repository holds 18 skills in this directory. The repository was last updated on July 23, 2026.

Source: protect-my-hair/nucleus-marketplace on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.