Skip to content

tim-hub/powerball-harness / .claude/skills/cc-update-review

cc-update-review

Verified

Reviews PRs that modify the Feature Table in CLAUDE.md. Use when a PR adds CC version rows or modifies docs/CLAUDE-feature-table.md.

CC Update Review Guardrail

A quality guardrail that prevents "doc-only" additions to the Feature Table during Claude Code update integration. Automatically classifies whether Feature Table additions are accompanied by implementation, and forces output of implementation proposals when lacking.

Quick Reference

This skill is triggered in the following situations:

  • When reviewing CC update integration PRs
  • When a diff is detected with new rows added to the Feature Table (CLAUDE.md / docs/CLAUDE-feature-table.md)
  • Internal invocation when /harness-review determines a PR is a CC integration PR

Situations where this skill is not triggered:

  • Normal implementation work (/work)
  • Changes only to files other than the Feature Table
  • Setup and initialization work

3-Category Classification

Each item added to the Feature Table is classified into the following 3 categories.

(A) Has Implementation

Definition: Implementation changes to hooks / scripts / agents / skills / core corresponding to the Feature Table addition are included in the same PR.

Criteria:

  • Files related to the feature mentioned in the Feature Table row have been changed
  • There is a diff in hooks.json, skill SKILL.md, agent .md, scripts/.sh, or core/src/.ts

Examples:

Feature Table AdditionCorresponding Implementation ChangeClassification
PostCompact hookhooks/post-compact-handler.sh newly createdA
MCP Elicitation supportElicitation event added to hooks.json + elicitation-handler.sh createdA
Worker maxTurns limitmaxTurns field added to agents/worker.mdA

Result: OK. No additional action needed.


(B) Doc-Only

Definition: Rows were added only to the Feature Table with no Harness-side implementation changes. Also does not qualify as CC auto-inherited (Category C).

Criteria:

  • New rows exist in the Feature Table
  • No related changes to hooks / scripts / agents / skills / core in the same PR
  • The feature is one where Harness should provide its own added value (configuration, workflow integration, guardrails, etc.)

Examples:

Feature Table AdditionCorresponding Implementation ChangeClassification
PreCompact hookNone (Feature Table only)B
Agent TeamsNone (Feature Table only)B
Desktop Scheduled TasksNone (Feature Table only)B

Result: NG. Block the PR and require an implementation proposal. Output format described below.


(C) CC Auto-Inherited

Definition: Items such as Claude Code core performance improvements, bug fixes, and internal optimizations that require no Harness-side changes.

Criteria:

  • A CC core fix with no room for Harness to wrap or extend
  • Performance improvements, memory leak fixes, UI improvements, etc.
  • Internal changes that do not affect Harness workflows

Examples:

Feature Table AdditionReasonClassification
Streaming API memory leak fixCC internal memory leak fix. No Harness-side action neededC
Compaction image retentionCC retains images during compaction. No Harness changes neededC
Parallel tool call fixCC internal parallel execution fix. Benefits are automaticC

Result: OK. However, "CC auto-inherited" must be explicitly noted in the Feature Table column.

CC Update PR Checklist

Verify the following in order during PR review:

## CC Update Integration Checklist

### 1. Extract Feature Table Diff
- [ ] List added rows from the diff of `CLAUDE.md` or `docs/CLAUDE-feature-table.md`

### 2. Classify Each Item
- [ ] Determine A / B / C for each added row
- [ ] Confirm that Category B items are 0

### 3. Per-Category Verification
- [ ] (A) Has implementation: Are corresponding implementation files correctly linked?
- [ ] (B) Doc-only: Has an implementation proposal been provided? (Block PR if not 0)
- [ ] (C) CC auto-inherited: Is "CC auto-inherited" explicitly noted in the Feature Table?

### 4. CHANGELOG Verification
- [ ] Are Category A items documented in the CHANGELOG in "Before / After" format?
- [ ] Are Category C items documented in the CHANGELOG as CC auto-inherited?

### Classification Results

| # | Feature Table Item | Category | Corresponding File / Notes |
|---|-------------------|----------|---------------------------|
| 1 | (item name) | A / B / C | (file path or notes) |
| 2 | (item name) | A / B / C | (file path or notes) |

Output Format When Category B Is Detected

When 1 or more Category B items are detected, output an implementation proposal in the following format. This output is mandatory and cannot be omitted.

## Category B Detected: Implementation Proposals

### B-{number}. {Feature Table item name}

**Current state**: Listed in Feature Table only. No Harness-side implementation.

**Harness-specific added value**:
{Specific explanation of how Harness should leverage this feature}

**Implementation proposal**:

| Target File | Change Description |
|------------|-------------------|
| `{file path}` | {specific change description} |
| `{file path}` | {specific change description} |

**User experience improvement**:
- Before: {current user experience}
- After: {user experience after implementation}

**Implementation priority**: {High / Medium / Low}
**Estimated effort**: {Small / Medium / Large}

Output Example

## Category B Detected: Implementation Proposals

### B-1. Desktop Scheduled Tasks

**Current state**: Listed in Feature Table only. No Harness-side implementation.

**Harness-specific added value**:
Integrate Scheduled Tasks with Harness workflows to automate periodic quality checks,
status syncing, and memory cleanup.

**Implementation proposal**:

| Target File | Change Description |
|------------|-------------------|
| `skills/harness-work/references/scheduled-tasks.md` | Scheduled task templates and guide |
| `scripts/setup-scheduled-tasks.sh` | Initial setup script |
| `hooks/hooks.json` | Cron trigger registration |

**User experience improvement**:
- Before: Users had to manually run periodic tasks
- After: Harness automatically runs periodic quality checks and notifies with results

**Implementation priority**: Medium
**Estimated effort**: Medium

Recommended "Added Value" Column

It is recommended to add the following column to the Feature Table:

FeatureSkillPurposeAdded Value
PostCompact hookhooksContext re-injectionA: Has implementation
Streaming leak fixallMemory leak fixC: CC auto-inherited

This column makes it possible to verify each item's classification at a glance and prevent Category B items from remaining.

Related Skills

  • harness-review - Code review (internally invokes this skill when a CC integration PR is detected)
  • harness-work - Implementation work (when working based on Category B implementation proposals)
  • memory - SSOT management (recording classification criteria decisions)
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 198kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 198kUpdated Claude CodeCodexVerifiedInstall