Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Review Gates

ASecurity

Multi-agent review-gate mechanics for repos shared with Codex and Antigravity — which gate lane a change requires, how to run and check a receipt, the handoff payload (claim to disprove + exact repro), review independence, and persisting verdicts. Use before running any review gate, before handing work to another agent, before merging, or when a gate refuses a receipt.

2 stars
0 votes
0 copies
0 views
Added 9/22/2026
ai-agentsgogit

Works with

terminal

Security Analysis

A100/100

Scanned 9/22/2026

Install to Claude Code

$npx -y skills add jckeen/dotfiles --skill review-gates --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Review Gates?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Review Gates
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/jckeen-review-gates/badge)](https://www.skillsdirectory.com/skills/jckeen-review-gates)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: review-gates
description: Multi-agent review-gate mechanics for repos shared with Codex and Antigravity — which gate lane a change requires, how to run and check a receipt, the handoff payload (claim to disprove + exact repro), review independence, and persisting verdicts. Use before running any review gate, before handing work to another agent, before merging, or when a gate refuses a receipt.
---

# Review gates and multi-agent handoffs

The always-loaded rules (one owner of the working tree, adversarial
verification, claim-to-disprove handoffs, verdicts as artifacts) live in the
global instructions. This skill carries the mechanics. Full role table and
rationale: `MULTI-AGENT.md` in the dotfiles repo.

## Assign roles explicitly

The active session is the conductor and owns planning, integration,
verification, delivery, and handoffs. Assign bounded implementation,
independent review, or runtime/browser verification to agents with the needed
capabilities. Any runtime can conduct or implement; personal defaults belong in
private preferences. Refute, don't rubber-stamp.

## One owner of the working tree

The Parallel agents rule applies across tools too. Each agent gets its own
worktree, or edits are sequenced. Never run two agents editing the same files
at once, and never share one git checkout between two interactive sessions.

## Verification is adversarial

Three agents agreeing can be one blind spot voted thrice. Assign the refuter
role explicitly; route disagreement to a fix, not a tie-break.

## Review independence

A fresh context reduces inherited assumptions; high-risk changes also require a
reviewer from a different model family. Record the actual reviewer evidence: a
requested model label alone does not establish the model used, and a text-diff
review is not browser evidence.

## Ask which lane before running a gate (ADR-0008)

```
review-receipt.py lane --repo . --scope committed
```

names the required lane.

- Ordinary tier-2 work goes to `antigravity-review-gate.sh`.
- A risk surface (or a classification the helper could not read) requires
  `codex-review-gate.sh` and is never downgradable.
- A tier-1 docs diff takes either gate's tier valve.
- `review-receipt.py check` refuses a receipt whose lane ranks below the
  requirement, so run the *named* gate rather than the familiar one, and check
  the receipt with no `--reviewer`.
- An Antigravity gate that exits 3 could not run and falls back to Codex; exit 2
  is a verdict and never falls back.
- `review-and-push.sh` performs the whole selection itself.

## Handoff payload

When handing work to another agent, the note carries the *claim to disprove*
and the *exact repro command*, not just "please review". For gate-mediated
refutation, pass them directly:

```
codex-review-gate.sh --claim "<claim>" --repro "<cmd>"
```

## Verdicts are artifacts

A rescue diagnosis or verification verdict from another agent must be persisted
(handoff note or issue comment) before the team acts on it — output that only
reached one terminal is lost work. Notes carry a "Session continuity" section
(codex session id / agy conversation id) so the next hand-back resumes instead
of cold-starting.

Attribution

jckeenjckeen
View sourceMore from jckeen →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1066601 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

651 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →