review-tribunal
Review tribunal for pull requests or local Git diffs: intake, review-core blind critics (including alternatives), tribunal, adversarial validation, then local…
Skill by lucavb
Chalk Labs
Need help deploying review-tribunal?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/lucavb-review-tribunal
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 26, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/review-tribunal
What's inside
- Skill name
review-tribunal- Description
- Review tribunal for pull requests or local Git diffs: intake, review-core blind critics (including alternatives), tribunal, adversarial validation, then local reporting or interactive GitHub posting. Requires review-core. Use when reviewing a PR, comparing a base branch with the current branch, reviewing a commit range or working tree, or spawning a review tribunal.
- License
- MIT
- Instructions
- 72 lines in SKILL.md · view on GitHub
Files (4)
- GITHUB-REVIEW.md
- LOCAL-DIFF.md
- README.md
- SKILL.md
SKILL.md (first lines)
# Review tribunal A **tribunal** always follows the same process: intake → **review-core** (blind → tribunal → adversarial) → delivery. The target selects the intake and delivery mechanics. **Requires [review-core](../review-core/SKILL.md)** installed as a sibling skill. When implementing from a ticket, use [ticket-to-pr-pipeline](../ticket-to-pr-pipeline/SKILL.md) Phases 3–4 with `mode: implement`. ## Steps ### 1. Select target and resolve diff Select one target before calling target-specific tools: - `github-pr` — the user supplied a PR URL/number or explicitly requested GitHub review. - `local-diff` — the user supplied a Git range, named a base and current branch, requested branch/commit/working-tree review, or invoked tribunal without a PR. If both targets are explicit, ask which one to review. For `github-pr`, fetch metadata, files, commits, existing reviews, and head SHA with `gh`; check out or fetch the PR branch so repository reads match the diff. For `local-diff`, follow [LOCAL-DIFF.md](LOCAL-DIFF.md). Its source manifest is the single source of truth for the reviewed range and working-tree scope. **Completion criterion:** exactly one target is selected; its diff is non-empty; the source manifest records the exact diff source, head SHA, changed files, and included working-tree state. ### 2. Gather intel 1. Fetch ticket/AC from Jira when identified, or use pasted AC. 2. Pull linked tickets and epic context when they change blast radius or ownership. 3. Ask for sibling-repo paths when the AC spans FE/libs. 4. Skim entrypoint routing, sibling implementations, and prior art on the base — facts only, not author intent. 5. When the target has commits, include commit SHAs in the source manifest. 6. If PR/ticket metadata references an upstream/downstream PR, shared event/API/schema, or dependency snapshot, read the linked contract and at least one consumer before making a semantic or compatibility finding; do not substitute ticket shorthand or PR rationale for the contract. **Completion criterion:** ticket/AC (or explicitly N/A), sibling paths (or N/A), linked sources inspected (or N/A with reason), and a factual intel brief are ready for critic prompts. ### 3–5. Review core (`mode: review`) Run [review-core](../review-core/SKILL.md) Steps 1–3 with **`mode: review`**. Inputs to pass: - Ticket + AC + linked tickets - Source manifest + diff - Intel brief from Step 2 - Sibling repo paths (local Read preferred) **Alternatives critic is required** — reviews must surface cleaner, simpler, or more consistent implementations, not just bugs. **Hygiene critic** when manifest lists commits — [review-core/COMMIT-HYGIENE.md](../review-core/COMMIT-HYGIENE.md). See [review-core/BLIND-REVIEW.md](../review-core/BLIND-REVIEW.md) and [review-core/ALTERNATIVES.md](../review-core/ALTERNATIVES.md). Publish scorecards to the user after tribunal and adversarial validation. Delivery begins only after the revised scorecard. **Completion criterion:** revised scorecard with adversarial verdicts on all Holds. ### 6. Deliver findings For `local-diff`, deliver the revised scorecard using [LOCAL-DIFF.md's local report](LOCAL-DIFF.md#local-report). For `github-pr`, walk remaining Holds and intentional Defer notes with the user using [GITHUB-REVIEW.md](GITHUB-REVIEW.md). That file is the single source of truth for pending reviews, post/continue, and submission. **Completion criterion:** local target — final evidence-linked report and audit trail delivered; GitHub target — pending review submitted or explicitly discarded, with posted/skipped summary. ## Checklist ``` - [ ] Target selected; exact diff source recorded - [ ] Ticket/AC + sibling paths + intel brief - [ ] review-core: blind critics (≥2 lenses; alternatives required; hygiene run or skip noted) - [ ] review-core: first tribunal scorecard - [ ] review-core: adversarial validation + revised scorecard - [ ] Target-specific delivery complete ```
From lucavb/skills, MIT.
More from lucavb/skills
- 0.604score
scrum-next
lucavb
Scrum board next-work triage.
SkillAFreshness: A★ 3Claude Code +9 - 0.601score
review-core
lucavb
Shared adversarial review engine: blind parallel critics (including alternatives), tribunal adjudication, and adversarial re-validation.
SkillAFreshness: A★ 3Claude Code +9 - 0.569score
refining-jira-tickets
lucavb
Pre-scrum refinement of Jira stories for engineering readiness.
SkillBFreshness: B★ 3Claude Code +9 - 0.569score
atlassian-jira-adf-formatting
lucavb
Round-trip Jira story descriptions via Atlassian MCP ADF — fetch full doc, edit in files, validate, file-backed editJiraIssue.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
conversation-handoff
lucavb
Summarize the current conversation into a copy-pasteable handoff message for a fresh chat.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
gh-fetch
lucavb
GitHub fetch — gh CLI for GitHub URLs; checkout when multiple file accesses are foreseen.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
closing-ritual
lucavb
Closing ritual — surface doubts, blind spots, and unverified claims before work is accepted.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
ticket-to-pr-pipeline
lucavb
ticket-to-pr-pipeline: Jira ticket through plan, build, review-core, and PR in one continuous run.
SkillBFreshness: B★ 3Claude Code +9 - 0.561score
writing-commit-messages
lucavb
Required before every git commit.
SkillBFreshness: B★ 3Claude Code +9 - 0.561score
angular
lucavb
Angular zoneless conventions — signals, toSignal, shells, encapsulated styles.
SkillBFreshness: B★ 3Claude Code +9 - 0.559score
discover-before-planning
lucavb
Do discovery during planning — explore the codebase and close open questions before presenting a plan.
SkillBFreshness: B★ 3Claude Code +9 - 0.558score
tldr
lucavb
Condense the immediately preceding assistant answer into a direct TL;DR.
SkillBFreshness: B★ 3Claude Code +9
Signals
- Stars
- 3
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 7, 2026 (26 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/lucavb/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from lucavb
- 0.604score
scrum-next
lucavb
Scrum board next-work triage.
SkillAFreshness: A★ 3Claude Code +9 - 0.601score
review-core
lucavb
Shared adversarial review engine: blind parallel critics (including alternatives), tribunal adjudication, and adversarial re-validation.
SkillAFreshness: A★ 3Claude Code +9 - 0.569score
refining-jira-tickets
lucavb
Pre-scrum refinement of Jira stories for engineering readiness.
SkillBFreshness: B★ 3Claude Code +9 - 0.569score
atlassian-jira-adf-formatting
lucavb
Round-trip Jira story descriptions via Atlassian MCP ADF — fetch full doc, edit in files, validate, file-backed editJiraIssue.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
conversation-handoff
lucavb
Summarize the current conversation into a copy-pasteable handoff message for a fresh chat.
SkillBFreshness: B★ 3Claude Code +9 - 0.564score
gh-fetch
lucavb
GitHub fetch — gh CLI for GitHub URLs; checkout when multiple file accesses are foreseen.
SkillBFreshness: B★ 3Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get review-tribunal running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
review-tribunal skill. Agent Mart. https://agentmart.dev/skills/lucavb-review-tribunal (last updated 2026-10-24)