Run Replay
Agent Run Replay Blackbox
Turn an AI-agent trace, browser steps, tool calls, console notes, screenshots, and outcome into a replayable timeline, JSONL sequence, postmortem, GitHub issue draft, and proof receipt.
BROWSER EVIDENCE EVAL LAB
Replay agent runs, verify citations, and build safe eval fixtures.
This kit produces evidence artifacts that humans and AI agents can inspect: timelines, JSONL replay files, claim-source matrices, heatmaps, synthetic fixtures, schema contracts, and proof receipts.
RAG Eval
Retrieval Citation Replay Studio
Turn AI answer claims, retrieved chunks, URLs, scores, and source notes into a claim-source matrix, unsupported-claim report, retrieval replay JSON, heatmap SVG, eval cases, and proof receipt.
Fixtures
Privacy-Safe Eval Fixture Builder
Turn a schema, task, field rules, edge cases, redaction policy, and seed into deterministic synthetic CSV, JSONL fixtures, schema contract, edge cases, PII risk report, and proof receipt.
ARTIFACT PROMISE
Every tool produces a replay, eval, or fixture package that can be reused.
Expected outputs include HTML timelines, JSONL replay files, CSV matrices, source heatmaps, schema contracts, synthetic data, markdown reports, and receipts.