# How do you reconstruct what your agents did overnight without reading every transcript? — retained record

> Retained reddit record from reddit:ClaudeAI.

Canonical: https://highsignal.app/data/records/af230f45d1ac3fe6

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeAI)  published 2026-10-29 16:30 ingested 2026-10-30 02:36

[original source ↗](https://www.reddit.com/r/ClaudeAI/comments/1xiotil/how_do_you_reconstruct_what_your_agents_did/)

## retained text



```
I start my agents at the keyboard and then leave them running overnight, usually across a few different projects. Some mornings there's a finished solution waiting. By the time I sit down I'm trying to piece together what each agent actually did, and reading transcripts is slow.

The git diff only gets me part of the way. It shows what changed but not whether an agent finished or got stuck waiting on a question.

What's helped a bit is the Kanban app I'm working on. Agents put their work on cards, so my morning check is reading cards instead of scrolling chats. When a feature also needs a policy update, the agent does the feature and puts up a separate card for the policy so it doesn't get forgotten. It doesn't fix everything, it just guides. Rules don't always fire, and the agent still has to write the card.

Does anyone else use a Kanban board for this? And what other tools do you use to see what your agents did overnight?

TLDR: I leave agents running overnight and used to lose the morning figuring out what they did. Cards help some. Anyone else using a board, or something different?
```



document fetched 2026-10-30 00:21 · canonical https://reddit.com/r/ClaudeAI/comments/1xiotil/how_do_you_reconstruct_what_your_agents_did
