opencode/picoclaw
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update AGENT_LOOP_IMPROVEMENTS.md

Browse source
This commit is contained in:
hobbyistlabs-coder 2026-03-13 18:40:09 -04:00 • committed by GitHub
parent 79c63b4837
commit 45cb8f8104
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
AGENT_LOOP_IMPROVEMENTS.md
View file

@ -16,7 +16,7 @@ This document outlines a series of tasks to improve the main loop of the agentic
* [x] **Graceful Recovery on Tool Panic:** Add `defer recover()` inside the parallel tool execution goroutines to prevent a panicked tool from crashing the entire `AgentLoop`. Return the panic as an error string to the LLM.
* [x] **Exponential Backoff:** Replace the linear backoff in LLM retries (`time.Duration(retry+1) * 5 * time.Second`) with exponential backoff and jitter to better handle rate limits.
* [ ] **Granular Error Classification:** Update `LLMProvider` interfaces to return structured, typed errors (e.g., `providers.ErrContextLengthExceeded`) instead of relying on fragile string matching.
* [x] **Granular Error Classification:** Update `LLMProvider` interfaces to return structured, typed errors (e.g., `providers.ErrContextLengthExceeded`) instead of relying on fragile string matching.
## Phase 3: Performance & Latency