Back to workTrace forensics · Private build

AgentRewind

Time-travel debugging for AI agent traces. Load a completed run and replay it step by step — cursor through decisions, tool calls, and state transitions after the fact instead of guessing from logs.

How it enforces trust

Trace files are treated as hostile input. The importer is zero-trust by construction — anything that fails a check is rejected before it reaches the replay engine.

Signed imports with JWT verification; unsigned or signature-mismatched payloads are refused.
Payload size double-check: declared size is verified against actual size before parsing.
MAX_STEPS clamp on replay length — no unbounded traces.
Input sanitization on every field before it hits the diff or replay path.
Five trace formats auto-detected on import — no manual format selection, no ambiguous parsing.

Diff engine

Two runs, side by side. The diff engine surfaces where behavior actually changed — not just where output text differs.

Divergence points: the first step where run A and run B disagree.
Latency deltas per step and across the full run.
Token-count deltas — input and output — for cost regression checks.

Why privateAgentRewind is a private build while it hardens toward release. No repo, no demo, no promised dates — the trust story has to be bulletproof before the parser touches anyone else's traces.

Read the proof. Then talk scope with TechTide AI.

This domain publishes implementation proof only. Commercial scope, deliverables, and pricing live on TechTide AI: AI Production Systems for agent and LLM work, Airtable Exit and Rescue for moving off Airtable.

AI Production Systems Airtable Exit and Rescue alex@techtideai.io

Also on TechTide AIProduction Block Call Critical System Implementation

alexcinovoj.dev