Quality Gate logo

Quality Gate

Community
athola
quality-gate

Orchestrates the QUALITY pipeline stage for egregore work items, running code review, unbloat, and test updates. Use when running quality checks before a PR.

Overview

Publisherathola
Repositoryclaude-night-market
Skill namequality-gate
Stars
343
Forks
37
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by athola on GitHub. Read the source before you install it.

Installation

Install the Quality Gate AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/athola/claude-night-market/tree/master/plugins/egregore/skills/quality-gate
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/athola/claude-night-market.git /tmp/claude-night-market
mkdir -p .claude/skills
cp -r /tmp/claude-night-market/plugins/egregore/skills/quality-gate .claude/skills/quality-gate
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Quality Gate in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Quality Gate on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Quality Gate is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Quality Gate

When To Use

  • Running quality checks on egregore work items
  • Self-review before creating a PR
  • Reviewing another agent's PR in PR-review mode

When NOT To Use

  • Manual code reviews outside egregore pipeline
  • One-off lint or format checks (use make lint directly)

Orchestrate the QUALITY stage of egregore's pipeline. Each quality step runs convention checks from the codex and invokes mapped skills.

Routing Table

StepConventionsSkillsModes
code-reviewC1,C2,C3,C4,C5pensive:unified-reviewself, pr
unbloat-conserve:unbloatself
code-refinement-pensive:code-refinementself
update-tests-sanctum:update-testsself
update-docsC5sanctum:update-docs, scribe:slop-detectorself

Inputs

The orchestrator invokes this skill with:

  • step: which quality step to run (e.g. "code-review")
  • mode: "self-review" or "pr-review"
  • work_item_id: the manifest work item ID
  • branch: the git branch with changes
  • pr_number: (PR-review mode only) the PR number

Self-Review Workflow

When mode is "self-review":

  1. Get changed files: git diff --name-only main...HEAD
  2. Load conventions from conventions/codex.yml
  3. Filter conventions to those mapped to the current step
  4. Run convention checks via conventions.py
  5. Invoke mapped skills on the changed files
  6. Collect all findings
  7. Calculate verdict

Auto-Fix Loop

If blocking findings exist:

  1. Attempt to fix each finding (skill-dependent)
  2. Commit fixes to the work item branch
  3. Re-run convention checks
  4. If still blocking after 3 attempts, verdict is "fix-required"

Verdict Calculation

if no findings:
    verdict = "pass"
elif all findings are severity "warning":
    verdict = "pass-with-warnings"
elif blocking findings remain after auto-fix:
    verdict = "fix-required"

Record verdict in manifest decisions:

json
{
  "step": "code-review",
  "chose": "pass-with-warnings",
  "why": "2 warnings (C4: noqa in hooks), 0 blocking"
}

Completion Integrity (opt-in, default OFF)

By default the loop runs indefinitely and autonomously: a fix-required verdict is recorded but does not, on its own, stop the item from advancing. This preserves the historical hands-off posture.

When config.pipeline.completion_integrity is true, the verdict becomes a gate the agent cannot talk its way past:

  1. A fix-required verdict is reported to the orchestrator as a step failure, so the item cannot advance to the ship stage with unresolved blocking findings. It retries in place and, on exhausting max_attempts, is marked failed and the overseer is alerted (never silently completed).
  2. Merge is held for human review regardless of auto_merge: the PR is prepared but left open.

This binds "done" to the verifier (convention checks plus the mapped review skills), not to the agent's own say-so. The principle comes from shipped agent harnesses, stated best by KbWen/agentic-os: an agent "can still cut a corner, it just can't cut this one past a check it doesn't control." The recurring patterns across those harnesses:

PatternWhat it doesSource
Verifier gate as completion judgeBind "done" to tests, lints, or result hashes the agent cannot fakeAider, snarktank/ralph, gaasher
Completion promise + max-iterations capString-matching "done" is insufficient; pair it with an iteration ceiling and manual abortanthropics ralph-wiggum
Evidence gates the agent does not controlCredential scan, test-execution validation, phase-log parsing enforced by hooks or CIKbWen/agentic-os
Keep-if-better / revert-if-worse ledgerScore each change, revert regressions, log to an append-only ledger for reviewgaasher/Agent-Loop-Skills
Legibility anchorA persisted current-state or repo-map file that lets a human re-enter the loopAider repo map, agentic-os
Human observer checkpointWatch early iterations, monitor diffs, reserve loops for defined problemsanthropics, ghuntley

Further rationale: make the illegal state unrepresentable before guarding against it, because a fallback that hides an impossible case becomes silent corruption instead of a locatable crash. Armin Ronacher, "The Coming Loop" (2026), https://lucumr.pocoo.org/2026/6/23/the-coming-loop/. The flag is off by default; enabling it is a deliberate choice to keep a human as the final judge.

PR-Review Workflow

When mode is "pr-review":

  1. Fetch PR diff: gh pr diff <number> --name-only
  2. Load conventions and filter to code-review step
  3. Run convention checks on changed files
  4. Invoke pensive:unified-review on the diff
  5. Collect all findings

Posting Reviews

Map findings to GitHub review:

  • No findings: gh api POST review with event "APPROVE"
  • Warnings only: POST review with event "COMMENT", findings as inline comments
  • Blocking findings: POST review with event "REQUEST_CHANGES", blocking findings as inline comments with "must fix" prefix

Comment format per finding:

[egregore:{convention_id}] {message}

Convention: {convention_name}
Severity: {severity}

Quality Config

Work items may have a quality_config field:

json
{
  "skip": ["unbloat"],
  "only": ["code-review", "update-docs"]
}
  • skip: list of steps to skip (run all others)
  • only: list of steps to run (skip all others)
  • If both are set, only takes precedence
  • If neither is set, all steps run (default)

Convention Filtering by Step

Not all conventions run on every step. The routing table above defines which conventions apply to which step. The quality gate filters the loaded codex accordingly before running checks.

Exit Criteria

  • All applicable convention checks executed
  • All mapped skills invoked
  • Verdict calculated and recorded in manifest
  • For PR-review: GitHub review posted

Frequently asked questions

What does the Quality Gate AI skill do?

Orchestrates the QUALITY pipeline stage for egregore work items, running code review, unbloat, and test updates. Use when running quality checks before a PR.

Why use Quality Gate on TypingMind?

Because you install it once and use it with any model. Quality Gate is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Quality Gate in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/athola/claude-night-market/tree/master/plugins/egregore/skills/quality-gate. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Quality Gate?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Quality Gate?

As many as you like. As long as a model supports skills, you can use Quality Gate with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Quality Gate AI skill free?

Yes. It is published on GitHub by athola under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇