Agent Run Replay & Review

Turn an agent run's prompts, tool calls, and results into a replayable timeline for review and sharing

Problem

When teams collaborate with coding agents, code review or bug triage often requires seeing what a colleague and the AI actually discussed, what context was given, and whether a critical edge case was considered. Today that means digging through chat logs or asking for a retelling, and context is usually lost.

Solution

A lightweight CLI / local service that hooks into common coding agents, records every event of a run (prompts, tool calls, file diffs, results), and produces a replayable static timeline plus shareable links. It supports search by session or issue so teams can discuss a specific step.

Users

Engineering teams using Claude Code / Codex, tech leads, code reviewers, and remote teams.

MVP

Capture Claude Code / Codex runs; structure into an event timeline; local viewer and static export; optional anonymous share link; keyword search across past sessions.

Revenue

Free open-source local version; hosted team tier priced per seat with share links, search, and retention policies.

Why now

Coding agents are now daily team tools, yet AI conversation context has no capture or sharing mechanism. As AI accelerates output, human coordination matters more than ever.