OMK

v0.78.3 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 1mo AI & Machine Learning

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

agent-orchestration agent-runtime ai-coding claude-code cli code-review

+14 more

codex coding-agent developer-tools devtools evidence local-first local-llm mcp model-context-protocol multi-agent opencode software-engineering typescript workflow-automation

Summary

AI summary

Updates v0.78.3, Design, and Tests across a mixed release.

Full changelog

v0.78.3 (2026-06-09)

Added

Provider Health Profiler v2 — capability vector state machine (missing → installed → auth_present → auth_valid → model_available → quota_available → sandbox_supported → tool_contract_verified → ready)
Evidence Trust Score v2 — 5-factor evidence quality scoring with penalties (reproducibility, independence, coverage_relevance, provenance_integrity, freshness)
Policy/Sandbox Enforcement v2 — capability lattice with conservative minByAuthority policy combination and enforcement proof
Merge Arbiter — patch scoring, conflict detection, winner selection with structured rationale
Replay Benchmark Suite — 10 task categories, 9 metrics, shadow mode for router v1/v2 comparison, CI integration

Design

Runtime Router v2 — constrained contextual bandit (LinUCB) architecture doc with shadow mode plan
DAG Planner v2 — typed ActionAtom compiler + conflict-aware scheduler design doc