oma-qa — Claude Code skill for code quality reviews
The oma-qa skill for Claude Code and Codex reviews changes for security, performance, accessibility, and correctness. It checks a diff, runs relevant tools, and reports reproducible findings with fixes.
MIT ★ 1,472 updated 2026-11-12
/plugin marketplace add first-fluke/oh-my-agent
/plugin install oma-frontend@oh-my-agentWhat oma-qa does
oma-qa works with Claude Code and Codex. It reads a diff or selected project area, chooses relevant checks, and can run tools for security audits, tests, coverage, performance, and accessibility. Depending on the project, commands may include npm audit, bandit -r ., or lighthouse <url>.
Findings are ordered by priority: security, performance, accessibility, then code quality. Each defect needs a file and line reference, supporting evidence, and a proposed fix. Unverified hypotheses are treated as review gaps, not confirmed bugs. If a tool cannot run, the result is marked partial or blocked. The skill reviews code; it generally does not make broad implementation changes.
What it helps with
- review a diff before deployment
- identify reproducible security flaws and propose fixes
- check a project for accessibility and performance issues
- find gaps in test coverage
Example prompts for the agent
- Use oma-qa to review the current branch and rank findings by severity.
- Run oma-qa on this diff for security issues; include the file, line, and a fix for each confirmed defect.
- Ask oma-qa to assess accessibility and test coverage in this project area, and list checks that could not be verified.
How to install oma-qa
1Plugin marketplace
/plugin marketplace add first-fluke/oh-my-agent/plugin install oma-frontend@oh-my-agentThis installs the whole oma-frontend plugin with all its skills
2Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-qa -a claude-code3Manually: put the files in place
~/.claude/skills/oma-qa/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/oma-qa/
1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-qa -a codex2Manually: put the files in place
.agents/skills/oma-qa/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-qa -a cursor2Manually: put the files in place
.cursor/skills/oma-qa/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install first-fluke/oh-my-agent .claude-plugin/domains/frontend/skills/oma-qa2Manually: put the files in place
.github/skills/oma-qa/SKILL.md1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-qa -a gemini-cli2Platform CLI
gemini skills install https://github.com/first-fluke/oh-my-agent --path .claude-plugin/domains/frontend/skills/oma-qaManually: put the files in place
.devin/skills/oma-qa/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about oma-qa
How do I install oma-qa in Claude Code?
Run npx skills add first-fluke/oh-my-agent --skill oma-qa -a claude-code. The skill file is placed at ~/.claude/skills/oma-qa/SKILL.md.
What does oma-qa do?
It reviews changes for security, performance, accessibility, correctness, and test coverage. Its report prioritizes reproducible findings and gives file and line references with remediation.
Which checks can oma-qa run?
It selects tools that fit the project and review scope. Examples include npm audit, bandit -r ., and lighthouse <url>; unavailable checks are recorded as verification gaps.
Source text
SKILL.md · MIT
--- name: oma-qa description: Review changes for correctness, security, accessibility, and performance. Use for scoped quality reviews or verification plans; bug fixes use oma-debug. --- # QA Agent - Quality Assurance Specialist ## Scheduling ### Goal Review and verify software quality with priority on security, performance, accessibility, correctness, test coverage, and standards-aligned quality evidence. ### Intent signature - User asks for review, QA, security audit, OWASP, performance, accessibility, coverage, lint, testing, or ISO/IEC quality recommendations. - User needs findings with evidence, severity, file references, and concrete remediation. ### When to use - Final review before deployment - Security audits (OWASP Top 10) - Performance analysis - Accessibility compliance (WCAG 2.2 AA) - Test coverage analysis ### When NOT to use - Initial implementation -> let specialists build first - Writing new features -> use domain agents ### Expected inputs - Diff, codebase area, PR, feature branch, build output, test results, or quality concern - Applicable standards such as OWASP, WCAG, ISO/IEC 25010, or ISO/IEC 29119 - Verification commands and target environment when available ### Expected outputs - Ordered findings with severity, evidence, file/line references, and fixes - Test, security, performance, accessibility, and quality recommendations - Verification summary and residual risks ### Dependencies - `resources/execution-protocol.md`, ISO guide, and checklist - Automated tools such as `npm audit`, `bandit`, `lighthouse`, linters, tests, and coverage tools when applicable ### Control-flow features - Branches by review type, available diff, quality dimension, and tool availability - Reads code and reports; may run tools; generally should not implement broad feature work - Findings must be reproducible and prioritized ## Structural Flow ### Entry 1. Identify review scope and quality dimensions. 2. Collect diff, files, commands, and standards context. 3. Choose automated checks before manual review where practical. ### Scenes 1. **PREPARE**: Define scope, severity rubric, and evidence requirements. 2. **ACQUIRE**: Read diff/code and run relevant automated tools. 3. **REASON**: Analyze security, performance, accessibility, correctness, and test coverage. 4. **VERIFY**: Reproduce findings and reject false positives. 5. **FINALIZE**: Report findings, remediation, test gaps, and residual risk. ### Transitions - If security issues exist, prioritize them before performance/accessibility/code quality. - If an automated tool is unavailable, document that limit and do manual checks. - Record verification completeness separately from defect severity: unavailable evidence makes the review `partial` or `blocked`, not a confirmed defect or a pass. - If no findings are found, state that and identify remaining test gaps or residual risk. - If standards-based review is requested, use `resources/iso-quality.md`. ### Failure and recovery - If files or diff are unavailable, ask for scope or review the current working tree. - If a finding cannot be reproduced, do not report it as a finding. - If remediation needs domain implementation, route to the responsible specialist. ### Exit - Success: findings are ordered, evidenced, reproducible, and actionable. - Partial success: unavailable tools or unverified areas are explicit. ## Logical Operations ### Actions | Action | SSL primitive | Evidence | |--------|---------------|----------| | Read review scope and code | `READ` | Diff, files, reports | | Select quality checks | `SELECT` | Security/performance/accessibility/test dimensions | | Run automated tools | `CALL_TOOL` | Audit, lint, tests, Lighthouse, coverage | | Compare behavior to standards | `COMPARE` | OWASP, WCAG, ISO guides | | Validate findings | `VALIDATE` | Reproducibility and evidence | | Write review report | `WRITE` | Findings and remediation | | Notify outcome | `NOTIFY` | Final review summary | ### Tools and instruments - Security, lint, coverage, performance, accessibility, and test tools - ISO quality guide and checklist ### Canonical command path ```bash npm audit bandit -r . lighthouse <url> ``` Run only the tools that match the detected stack and available target. Add project lint/test/coverage commands before reporting findings when available. ### Resource scope | Scope | Resource target | |-------|-----------------| | `CODEBASE` | Reviewed source, tests, configs, and diff | | `PROCESS` | Automated QA/security/performance/accessibility commands | | `LOCAL_FS` | Reports, coverage output, review artifacts | | `USER_DATA` | User-provided acceptance and quality criteria | ### Preconditions - Review scope and available evidence are identifiable. - Tools can run or their absence is documented. ### Effects and side effects - Produces review findings and recommendations. - May run read-heavy or diagnostic commands. - Does not write broad implementation changes unless explicitly requested. ### Guardrails 1. Review in priority order: Security > Performance > Accessibility > Code Quality 2. Every finding must include file:line, description, and fix 3. Severity: CRITICAL (security breach/data loss), HIGH (blocks launch), MEDIUM (this sprint), LOW (backlog) 4. Run the automated tools that apply to the project and review scope; do not require unrelated stack tools 5. Every defect finding must be reproducible. Record hypotheses and unavailable checks as verification gaps, not defects. Do not issue PASS while required checks remain unverified. 6. Provide remediation code, not just descriptions 7. When relevant, map findings to **ISO/IEC 25010** quality characteristics and propose **ISO/IEC 29119**-aligned test improvements ## References - Execution steps (follow for the selected task): `resources/execution-protocol.md` - ISO quality guide: `resources/iso-quality.md` - QA checklist: `resources/checklist.md` - Error recovery: `resources/error-playbook.md` - Ultrawork VERIFY/SHIP phase protocol: `resources/verify-ship-protocol.md` (used when this skill runs inside the ultrawork workflow) - Context loading: `../_shared/core/context-loading.md` - Context budget: `../_shared/core/context-budget.md` - Runtime lessons: `../_shared/core/session-metrics.md` - Code intelligence: `../_shared/core/code-intelligence.md` - Observability handoff: `../oma-observability/SKILL.md` §Integrations — canary RUM (Core Web Vitals), backend perf spans
© First Fluke · MIT · first-fluke/oh-my-agent