| name | pos-qa |
| description | QA composite — browser testing, health scoring, and evaluation with test result memory. Replaces 5 QA skills. |
| argument-hint | [test|audit|browse] [url or path] |
pos-qa
QA composite — browser testing, health scoring, and evaluation with test result memory. Replaces 5 QA skills.
Replaces: qa, qa-only, browse, agentic-eval, self-eval
Actions
| Action | What |
|---|
test [url] | Full QA — browser testing, health score 0-100, bug detection |
audit [path] | Code quality audit without browser (static analysis + tests) |
browse [url] | Headless browser interaction — navigate, screenshot, inspect |
Routing
- Parse action from first argument
- Load domain memory from
~/.productionos/domains/pos-qa/
- If first run: auto-detect project context, create profile
- Dispatch to
sub-skills/{action}.md
- Score against
evaluation/rubric.yml, update memory
Domain Memory
Stored at ~/.productionos/domains/pos-qa/:
test-history.jsonl (health scores over time), known-bugs.jsonl (tracked issues with status), baseline-screenshots/ (visual regression reference), learnings.jsonl
Inputs
| Parameter | Values | Default | Description |
|---|
action | test, audit, browse | test | Sub-skill to run |
target | path, URL, or description | required | What to operate on |
Guardrails
- Memory is append-only. Never delete history.
- Every action scored against rubric. No silent completion.
- Profile is user-controlled. Never change without asking.
- Evidence required. No fabricated metrics.
- Learn from history. Compare to past executions.
