accessibility
Audit HTML and templates for WCAG 2.1 AA accessibility violations (labels, alt text, focus management, semantic structure).
- Capability
- read-only
- Requires
- None declared
Factory documentation / Stations
A browsable reference for the factory’s declared agents: class, execution plane, containment ceiling and requirements.
Each card shows the declaration, not a promise that every host grants the capability. Actual execution is checked at runtime. Chips read class, then execution plane, then containment tier; the coloured top edge repeats the class.
Audit HTML and templates for WCAG 2.1 AA accessibility violations (labels, alt text, focus management, semantic structure).
Measure bundle size, track size deltas vs baselines, and propose tree-shaking, code splitting, and byte budgets.
Audit dependency vulnerabilities, assess exploitability and reachability, and detect supply chain risks.
Detect drift between documentation and actual repository implementation.
Companion to docs-drift that synthesizes concrete markdown patches and PR updates to fix stale documentation and undocumented modules.
Triage incoming issues, verify reproduction steps, detect duplicates, and propose labels and actions.
Inspect runtime, test, and console logs, detect error signatures, correlate with source, and propose fixes.
Detect heap bloat and memory leak vectors (uncleaned listeners, unbounded caches, detached DOM nodes, timers) wrapping memory-leak-debugging + DevTools MCP.
Audit HTML, CSS, and JS against Modern Web Guidance to replace legacy workarounds with native Baseline Web APIs.
Goal-directed Class C optimizer that iteratively measures metrics, proposes bounded optimizations, verifies gains, and records dead ends in an experiment ledger.
Inspect recent git commits and diffs for performance anti-patterns (LCP, INP, CLS, layout thrashing, waterfalls) and propose concrete code fixes.
Synthesize minimal, verified code patches and PR diff proposals for failing CI checks or verified factory findings.
Meta-agent QA station that audits the precision, wontfix suppression rates, noise creep, and contract compliance of all other Factory agents.
Synthesize clean, high-impact release notes from git commits and PRs grouped by category.
Audit web applications against the 46 failure states (offline, timeout, storage quota, FOIT, 3P SPOFs, tab discard) wrapping web-resilience-plugin.
Find committed credentials, high-entropy secrets, and sensitive tokens.
Identify untested modules and functions, measure test deficit, and propose test suites.
Bootstrap and maintain THREAT_MODEL.md from code, git history, and issue tracking.
Analyze UI & UX quality—design token consistency, interactive focus/active/disabled states, loading/empty/error states, tap targets, dark mode, and visual hierarchy.
Surface vulnerability hypotheses and exploit chains across untrusted attack surfaces guided by repository threat models.
Deduplicate findings by root cause, calibrate against THREAT_MODEL.md, and rank reachability.
Adversarial verifier that independently stress-tests vulnerability candidates to disprove false positives.