feat(dashboard): markdown rendering, live announcements, model transparency

[gitjaeger2017]

• Render artifact output as real Markdown (headings, lists, tables, code fences, links) via a dependency-free DOM-building renderer that never uses innerHTML on model output; safe-URL check on links.
• Add an aria-live region announcing real-time stage/run transitions for screen-reader users, plus role=status on the gate banner.
• Persist run history to a run.json sidecar per run, reloaded at webapp startup so history survives a restart; any run interrupted mid-stage is marked as an error on reload instead of showing stuck "Running...".
• Surface which Fireworks model ran each stage; link to Fireworks' live pricing page instead of guessing at a cost figure.
• Add an opt-in browser notification + beep on run completion when the tab isn't focused.
• Fix: stage-card__output missing its display:none when hidden (an earlier markdown.css rule was out-ranking the browser's default [hidden] behavior), so "Hide output" now actually collapses the card.

[gitjaeger2017]

Transparency will be a huge grade for us as long as its seemless.

[graphify-labs]

Looks safe to merge — no coupling regressions and no blocking issues, checked against the code graph (not a self-assessment).

Graphify reviewed this change.

Looks safe to merge — no coupling regressions and no blocking issues, checked against the code graph (not a self-assessment).

---

Graphify review — findings

This pull request removes the Exploratory QA agent along with two research specialist agents (technology-scout and benchmarking) and one challenge-division agent (engineering-cross-domain-review), deleting their spec files and removing corresponding entries from the various README rosters. It also reverts the Validator agent's mission text and version to exclude Exploratory QA reconciliation, and simplifies the pipeline diagrams and stage descriptions in the top-level README accordingly. The associated changes appear to extend into orchestrator source, frontend, and test files to reflect the revised set of pipeline stages.

No blocking issues surfaced. 4 lower-confidence candidates did not survive cross-model review.

Analysis details — impact, health, verification

Impact & health

Graphify review

Impact — 199 functions depend on the 193 functions this change touches.

Health — this change adds coupling hotspots:

• worse: applyEvent() — 1 callers, 7 callees
• worse: applySnapshot() — 1 callers, 7 callees

Verification — 199 functions in the blast radius were not formally verified this run (proofs are advisory here).

Gate & verification

graphify gate

PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.

Advisory (not blocking):

• verification_scope: 199 function(s) in the blast radius were not formally verified this run

· 1 grounded finding(s) anchored inline below; 1 more finding(s) on lines outside this diff (see the check run).