Skip to content

Port RLM observation feedback to the generated runtimes - #914

Merged
dosco merged 5 commits into
mainfrom
codex/clear-axir-rlm-parity
Nov 18, 2026
Merged

dosco merged 5 commits into
mainfrom
codex/clear-axir-rlm-parity

Conversation

@dosco

@dosco dosco commented Nov 18, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Make the Go, Python, Rust, Java, and C++ embedded JavaScript runtimes pause for tool discovery and require a fresh model turn before acting on inspection results. Keep initialized values available across the pause.
  • Allow explicitly deterministic tool results to chain within one turn. Independent batches load every missing contract before dispatch and return ordered outcomes, including partial failures. The embedded adapters dispatch batch members through synchronous host callbacks.
  • Keep the latest feedback visible, omit paused source from replay, compact older observations, and reuse checkpoints across short stretches of new activity.
  • Regenerate all five packages and complete the RLM feedback entry in the AxIR backlog.

Verification

  • Each of the five real-engine suites passed all 17 cases. The inspection regression fails on the previous main because the request omits the observation.
  • The TypeScript-derived replay case, AxIR semantic parity, 44 focused TypeScript tests, 72 fast compiler tests, and skill validation passed.
  • The generated Go and Python runners passed all 260 agent conformance cases each. The generated Go package tests also passed with loopback access.

@dosco
dosco merged commit ca599e5 into main Nov 18, 2026
27 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant