test-opencode-tooling

To diagnose tool-call failures, run automated paired testing against a running afm instance to surface errors and compile a structured diagnostic report highlighting root causes without modifying runtime behavior.

Automated tool-call reliability testing runs OpenCode tasks against afm to surface generation or translation errors in a reproducible loop, validating new model integrations and parser updates across prompts.

Yes, you need a running OpenCode server and an active afm instance. The Skill requires a traceable tool-call pipeline to distinguish afm translation bugs from model generation errors.

Diagnostic reporting classifies model generation errors and afm translation bugs to guide debugging focus, but it does not apply fixes. It compiles structured results suggesting next steps without modifying runtime behavior.

Cross-model reliability experiments often surface tool-call failures due to afm translation bugs or model generation errors. Running automated paired testing identifies and classifies these specific failure points for debugging.