Millrace

v0.18.2 Feature

This release adds 2 notable features for engineering teams evaluating rollout.

Published 2mo AI Agents & Assistants

✓ No known CVEs patched

✓ No known CVEs patched in this version

Topics

agent-loop agentic-ai-development agentic-coding agentic-engineering agentic-framework agentic-workflow

+7 more

autonomous-agents governed-autonomy governed-loop harness-engineering loop-engineering runtime-engine runtime-engineering

Summary

AI summary

Added an opt-in Integrator execution stage with new Codex run modes and JSON formatting for status commands.

Full changelog

Added an opt-in Integrator execution stage and execution.with_integrator
graph loop for high-assurance Codex runs. The new
default_codex_integrated and learning_codex_integrated modes run
builder -> integrator -> checker, with Integrator writing
integration_report.md before normal Checker QA.
millrace status and millrace status show now support --format json,
including blocked-idle diagnostics and the latest runtime error report path.

Invalid Recon handoff artifacts now block the active probe with
recon_handoff_invalid instead of letting malformed probe output fall into
ordinary planning recovery.
Graph-loop asset validation now rejects Recon handoff outcomes wired directly
to stage nodes; typed Recon promotion must remain runtime-owned.
Stage-request construction now refuses stale stage/work-item pairings before
runner invocation, preventing Manager from receiving probe work.

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Track Millrace

Get notified when new releases ship.

About Millrace