AgentPlayback finds Claude Code spent 48 hours a week just waiting for its human
ruididnothingwrong · reddit · 2026-09-07
A developer open-sourced AgentPlayback, a tool that parses local AI coding session logs:
- Headline finding: over one week, his Claude Code / Codex sessions accumulated 48 hours of "waiting for human input" — waiting on him is a full-time job
- The tool builds a circular timeline visualizing working, waiting-for-input, and overlapping sessions across projects, all running locally
- UI was hand-designed in Figma; implementation assisted by the Fable model
- Free and open source: try it with npx agentplayback
A practical way to quantify the human-in-the-loop bottleneck in agent workflows.
More from coding & agent
- LangChain ships Managed Deep Agents 0.9, walkthrough from Interrupt London keynote — LangChain · 2026-10-26
- Knowledge Worker Kit adds a full section on organizing projects — dSebastien · 2026-10-26
- Spring AI 2.1.0-RC1 adds tool-call confirmation, pausing agent loops before real-data access — therealdanvega · 2026-10-26
- METR red-teams OpenAI's agent monitoring, finds bypasses and gaps — thione · 2026-10-26
- OpenAI: coding agents weakened safety monitoring, one lied about it — thione · 2026-10-26
- ConvFill: A 0.6B Talker Buys Time While a Large Reasoner Whispers the Real Answer — andimarafioti · 2026-10-26