Where should agent tool-call state live: model context or app?
LowMixture1084 · reddit · 2026-10-02
A developer's design dilemma in a small agent app:
- Model-owned state keeps prompts simpler, but retries and partial failures get weird fast.
- App-owned state gives cleaner control flow, but the agent starts feeling like a scripted workflow.
- The recurring failure mode: duplicated or stale tool calls after timeouts, especially when one tool mutates state and the next assumes it didn't.
More from coding & agent
- Microsoft Makes WSL Containers Generally Available for Running Linux Containers on Windows — usamawahabkhan · 2026-10-02
- Grok Build Adds Agent Dashboard to Manage All Your Agents From One Screen — RachelVT42 · 2026-10-02
- WP Astro MCP automates WordPress-to-Astro migration via Claude Code — modelcontextprotocol · 2026-10-02
- ThinAir Geo brings geocoding and routing intelligence to AI agents — modelcontextprotocol · 2026-10-02
- Modal launches BYOC 2.0 to keep cloud coding agents' sensitive data inside your account — charles_irl · 2026-10-02
- Dev Uses AI to Build Addictive Three.js Shooter Canyon/Overdrive for Web and Mobile — AIandDesign · 2026-10-02