lsp-chat-recovery-loop

name	lsp-chat-recovery-loop
description	Template workflow for diagnosing LSP chat-delivery regressions when pyagent introduces an LSP/runtime pipeline.

LSP Chat Recovery Loop (pyagent template)

This skill was adapted from another repo and is intentionally reduced to a template.

Current repo status

• pyagent currently has no LSP chat runtime, no e2e/ harness, and no log simplifier scripts.

When to activate this workflow

Use this only after the repository adds:

• an LSP/chat runtime,
• reproducible e2e scenarios,
• structured server/client logs.

Minimal loop (future)

1. Reproduce with a deterministic scenario.
2. Capture server/client logs for the run.
3. Quantify missing pipeline stages.
4. Write one falsifiable hypothesis.
5. Implement one minimal fix.
6. Re-run scenario and compare markers.
7. Document verification commands and expected evidence.

Rule

Keep all run artifacts in .scratch/projects/<project>/runs/<timestamp>/.