# Agent Talk

> Agent Talk (agentalk.org) is a field manual for people who already run AI agents and want them to finish real work. It publishes setup guides, failure post-mortems and operating cases, with a focus on running many agents in one codebase: handoffs, shared state, concurrent edits, review and verification loops, deploy safety, and cost and model routing. It also keeps a case library of open-source agent builds by use case (what each solves, how to run it, what was observed, link to the code) and side-by-side framework comparisons. Tool behaviour is quoted from each tool's own documentation; operating rules come from daily first-hand use.

## Tracks
- [Setup](https://agentalk.org/topics/setup/)
- [Multi-agent coordination](https://agentalk.org/topics/coordination/)
- [Review & deploy safety](https://agentalk.org/topics/verification/)
- [Cost & model routing](https://agentalk.org/topics/cost-routing/)
- [Post-mortems](https://agentalk.org/topics/post-mortems/)
- [Operating cases](https://agentalk.org/topics/operations/)
- [Agent case library](https://agentalk.org/topics/case-library/)
- [Frameworks compared](https://agentalk.org/topics/frameworks/)

## About
- [About](https://agentalk.org/about/)
- [Disclosure](https://agentalk.org/disclosure/)

## Guides