AI Execution Forensics — see what the AI actually did.
npm install
npm run devOpen http://localhost:3000. Requires a running deeplossless instance.
| Page | Description |
|---|---|
/ |
Signal Trace — classified event timeline with burst folding. Findings sidebar with confidence/source labels. Share card export |
/replay/[id] |
Execution Cinema — narrative execution analysis with stats, signal trace, evidence chain |
/plan/[id] |
Execution Divergence — plan vs actual comparison |
/health/[id] |
Execution Corruption — DAG anomalies as runtime diagnostics |
/stability |
Cache Stability — prompt hash heatmap + system prompt diff viewer |
/latency |
Latency Dashboard — canvas sparkline, P50/P95/P99 stats, recent records |
Next.js 16 · Tailwind CSS 4 · shadcn/ui · Motion · html-to-image
src/
app/ Pages (/, /replay, /plan, /health, /stability, /latency)
components/ signal-trace, claim-evidence, diff-evidence, share-card, status-bar
lib/
api.ts API client (auto-refresh every 5s)
classify.ts Shared tool → semantic category classification
rule-engine.ts Integrity detection (confidence/source labels)
types.ts Shared TypeScript types
use-sessions.ts Session data hook with auto-refresh