agent-audit
Audit agents spawned in the current/last run against the agent-selection taxonomy
Progressive multimedia enrichment pass — add high-value audio/video/image/interactive to a site, run again and again
> /plugin marketplace add heymegabyte/claude-skillsHow it fires
How this command gets triggered: by you, by Claude, or both.
/multimedia-enrichContext preview
What this command does when you run it.
Progressive multimedia enrichment pass — add high-value audio/video/image/interactive to a site, run again and again
description: Progressive multimedia enrichment pass — add high-value audio/video/image/interactive to a site, run again and again argument-hint: [route or "whole site"]
Run a repeatable progressive-multimedia enrichment pass per [[website-build-doctrine]] (Phase 2 maximalist enrichment + Phase 4 AI-native spiral already own podcasts/Veo/maps — this command is the iterative discipline that ships them page-by-page).
**Purpose** — make a website measurably richer every run: the right audio/video/image/interactive media, added where it helps the user, never at the cost of performance or a11y.
**When to use** — after a site is functional and you want to deepen it; re-run until satisfied. Targets `$ARGUMENTS` (a route, or `whole site`).
**Inputs** — `$ARGUMENTS` (route or `whole site`); the live project; `docs/multimedia-roadmap.md` if it exists (prior runs' backlog).
**Outputs** — new media sections + reusable media systems + updated `docs/multimedia-roadmap.md` with a prioritized backlog for the next run.
**Verification** — project's normal gates (lint/typecheck/unit/Playwright/build/preview/a11y/perf) green; media SSR-safe; no CWV regression per [[quality-metrics]]; deploy + prod-E2E per [[verification-loop]].
**Can update ~/.agentskills or ~/.claude?** NO — this enhances a project, not global config. Only `/self-improve` touches global.
1. **Inspect** the site (live + source); list every page + its current media. 2. **Find weakest pages** — thinnest, lowest-engagement, most generic surfaces first. 3. **Find UN-implemented opportunities** — never repeat an add already in the roadmap. 4. **Add highest-value first** — one earned win beats five filler embeds. 5. **Protect performance + a11y** at every add (lazy-load, reduced-motion, captions, alt). 6. **Update `docs/multimedia-roadmap.md`** — record what shipped + the next backlog. 7. **Leave a prioritized backlog** so the next run starts richer. Each run makes the site meaningfully better.
Fan out specialist media agents per [[agent-selection]] (audio-systems / video / image / motion) — never one generic worker for all four.
`MediaSection` · `PodcastPlaylist` · `AudioPlayer` · `AudioVisualizer` · `VideoFeature` · `ImageGallery` · `ImageLightbox` · `ResponsiveImage` · `AnimatedStats` · `Timeline` · `ShareCard` · `SeoMetadata` · `StructuredData` · `ReducedMotionProvider` · `MediaPerformanceGuard`.
## Page Multimedia Audit: <route> - **Current state** — what media exists now - **Best opportunities** — ranked by user value - **Add now** — highest-value, ship this run - **Wait** — deferred + why - **Why** — the user benefit each add serves
# Multimedia Roadmap ## Added this run ## Pages improved ## New components ## Placeholders (demo, need real media) ## Perf safeguards ## A11y safeguards ## Best next ## Deferred ## Risks
| Area | Result | Notes | |---|---|---| | Media improves context | | | | No pointless gimmicks | | | | Audio user-initiated | | | | Video lazy-loaded | | | | Images optimized | | | | Visualizer SSR-safe | | | | Reduced motion supported | | | | Accessibility covered | | | | CTAs included | | | | Roadmap updated | | |
Each row PASS/FAIL. Any FAIL → repair + rerun the project's checks. Never leave broken multimedia.
14-category autonomous product-building OS for 32+ AI coding tools. One-line prompts → deployed products.
Repo: heymegabyte/claude-skills
Audit agents spawned in the current/last run against the agent-selection taxonomy
Run the Agent Diversity Review gate and emit the result table
Meta-analyze the effectiveness of a /loop arc — per-iteration metrics, LOC delta trend, saturation detection, and a keep/lengthen/delete recommendation.
Audit the rules/ directory for missing foundational principles; output gap list with priority and justification
Validate ~/.claude/settings.json hooks block — event names, file existence, executability, matcher syntax; --fix repairs common issues
Catch Resend-class bug (isError: false on HTTP 4xx/5xx) across all MCP server tool handlers