brimyr-quality-review
After the Brimyr gate finishes on a GitHub PR, review the test and code quality its numbers cannot measure and triage the net-new quality findings it reported…
Skill by MagmaMoose
Chalk Labs
Need help deploying brimyr-quality-review?
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/magmamoose-brimyr-quality-review
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
Use cases: testing
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 Nov 2, 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/brimyr-quality-review
What's inside
- Skill name
brimyr-quality-review- Description
- After the Brimyr gate finishes on a GitHub PR, review the test and code quality its numbers cannot measure and triage the net-new quality findings it reported without blocking on, then reply on the PR with one idempotent comment carrying the additive findings.
- License
- Apache-2.0
- Instructions
- 99 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
Use the shared MagmaMoose post-Brimyr quality review workflow. Read and follow: - `shared/brimyr-quality-review.md` - The target repository's `CLAUDE.md` - The target repository's `AGENTS.md` - The target repository's `CONTRIBUTING.md` - Relevant `README.md` files Treat target-repository hard rules as blockers. Expected input: - PR number, PR URL, branch name, or enough context to identify the pull request. - Optionally, the Brimyr run that just finished. The workflow resolves it either way. ## What Brimyr leaves for you Brimyr is quality assurance and gates on two things: 1. **Patch coverage** — the fraction of the lines this PR changed that the test suite executed, against a threshold (80% by default, and not applied at all below `min_lines`). 2. **Net-new quality findings** — optional and off by default, and when on, **report-only out of the box**. It doesn't re-implement a linter: it calls `chargate filter-sarif` across a process boundary and gates on the counts. `quality_fail_on` defaults to `none`. Its outputs are `mode`, `gate_result` (`pass` | `fail` | `error`), `patch_coverage`, `covered_lines`, `total_lines`, `total_coverage`, `quality_gate_result`, `quality_net_new_count`, `quality_blocking_count`, `quality_fail_on`. It also runs `sonar-scanner` non-blocking for the quality trend when `sonar_url`, `sonar_token` and a `sonar-scanner` binary are all present (skipped with a reason otherwise), and posts one consolidated PR comment when `pr_comment: 'true'`. The coverage number says every changed line ran, not that anything was asserted. The finding count says findings exist, not that anyone judged them: with `quality_fail_on: none`, a green Brimyr routinely ships a list of real PR-scoped defects it blocked on none of. Those two gaps are this review's subject. ## Expected behavior - **Read Brimyr's PR comment first, and be honest about the source.** It is the primary source and carries both halves of the verdict. Find it by marker, never by author: `<!-- brimyr:pr-summary -->` is the consolidated comment (coverage plus, when quality is on, net-new findings); `<!-- brimyr:quality-summary -->` belongs only to a standalone `brimyr lint` run and is not the normal path. `pr_comment` defaults to `false`, so when there is no comment, fall back to the check run and then to the action outputs read off the workflow run. State in your comment which source you used, and never describe a comment you did not read. - **Review test quality**, the dimension coverage cannot measure: a test that executes a line without asserting on it, a test pinned to the implementation rather than the behaviour, an over-mocked test that would pass with the real dependency deleted, a test whose name and comment describe one contract while the assertion enshrines the opposite, missing negative and error-path cases, missing boundary cases, a flaky construction (time, ordering, network, randomness), and a test that cannot fail. For each one, name the mutation that should break the test and doesn't. - **Review code quality past the number**: duplication the diff introduces, a function doing two jobs, a name that lies, an abstraction with one caller, error handling that swallows, a lazy import that defers a dependency requirement rather than removing it, dead code, a comment that contradicts the code, and complexity you can tie to a named failure path. - **Judge the uncovered changed lines individually.** A gate at or above threshold can still leave the single most dangerous line unexecuted. An uncovered error path is worth more than ten covered getters. Note that Brimyr lists uncovered lines only when the gate *failed*, so on a pass you have to fetch the coverage report yourself. - **Triage the net-new quality findings, don't restate them.** With `quality_fail_on: none` a passing Brimyr routinely reports findings it did not block on, and nobody is reading them. Decide which matter and why the rest can wait; promote one only when you can name the concrete failure it causes. Severity there is the **SARIF level** (`error` / `warning` / `note`) — the producing linter's confidence, not a security band, so never map it onto your severity markers mechanically. And a scan that did not complete (`quality_gate_result: error`, the "did not complete" line, or the "these linters did not run" warning) is a tool error, never zero findings; an absent quality block means the half was never enabled. - **Deduplicate across three axes**: Brimyr's coverage output, Brimyr's own net-new findings list, and SonarQube — plus Chargate and any existing review thread. Say how many you dropped and against which sources. Never restate a gate number as a finding. - **Stay in your lane.** Brimyr's quality half calls `chargate filter-sarif`, so both gates emit Chargate-derived findings. The boundary is the **subject**, not the tool: security findings belong to `chargate-security-review`, quality findings to you, even when a security finding turns up inside Brimyr's listing. - **Post exactly one comment, idempotently.** It carries `<!-- agent-skills:brimyr-quality-review -->`; find the prior one and `PATCH` it, else `POST`. Link Brimyr's comment when there is one, and say the gate comment was absent when there isn't. Include the machine-readable findings block so `pr-triage` can act on it, and state plainly when nothing was found. - **Verify before reporting.** Every `file:line` real at the head commit, no invented markers or outputs, no padding. ## Hard rules - Never soften, downgrade, or drop a real finding because the gate came back green. A passing percentage is not evidence about any finding you hold, and a passing quality gate is weaker still: at the default `quality_fail_on: none` it passes by construction, so "Brimyr didn't block on it" is never an argument. - Never propose a fix whose only effect is to move the percentage. A test that calls the changed function and asserts nothing is a finding, not a remedy. - This workflow posts a comment. It does not submit a GitHub review, set a status check, or push code. `pr-triage` fixes what it finds. - No em-dashes in the posted comment, no robot emojis, no "AI review" branding, and never an attribution footer on anything posted. That punctuation rule matches `pr-review` and `chargate-security-review`, and applies to GitHub text only, not to this repo's files.
From MagmaMoose/agent-skills, Apache-2.0.
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Aug 24, 2026 (2 months ago)
- Last release
- Oct 23, 2026 (25 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- Apache-2.0
- Repository
- github.com/MagmaMoose/agent-skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
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 brimyr-quality-review 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
brimyr-quality-review skill. Agent Mart. https://agentmart.dev/skills/magmamoose-brimyr-quality-review (last updated 2026-10-24)