Skip to content

Keesan12/Martin-Loop

v0.2.10 Feature

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

Published 1mo AI Agents & Assistants
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

agent-runtime ai-agent-runtime ai-coding-agents ai-control-plane ai-governance ai-safety
+13 more
audit-trail budget-enforcement claude-code codex coding-agents control-plane governed-runtime llmops mcp model-context-protocol observability policy-as-code rollback

Affected surfaces

auth rbac

Summary

AI summary

Verifier receipts stay trustworthy and contradictions are surfaced as warnings.

Changes in this release

Feature Low

CLI and MCP verification views now include verifier warnings plus step‑level evidence.

CLI and MCP verification views now include verifier warnings plus step‑level evidence.

Source: llm_adapter@2026-06-07

Confidence: high

 —
Feature Low

Badge generation uses lightweight persisted‑run probe for readiness evidence to stay responsive on large run stores.

Badge generation uses lightweight persisted‑run probe for readiness evidence to stay responsive on large run stores.

Source: llm_adapter@2026-06-07

Confidence: high

 —
Bugfix Medium

`--runs-dir` flag consistent across CLI commands (doctor, start, preflight, run, dossier, badge).

`--runs-dir` flag consistent across CLI commands (doctor, start, preflight, run, dossier, badge).

Source: llm_adapter@2026-06-07

Confidence: high

 —
Bugfix Medium

Public help for `run --help` and `preflight --help` exits cleanly before governed-flow checks.

Public help for `run --help` and `preflight --help` exits cleanly before governed-flow checks.

Source: llm_adapter@2026-06-07

Confidence: high

 —
Bugfix Medium

Contradictions between adapter output and MartinLoop verifier are surfaced as warnings instead of hidden.

Contradictions between adapter output and MartinLoop verifier are surfaced as warnings instead of hidden.

Source: llm_adapter@2026-06-07

Confidence: high

 —
Bugfix Medium

`verification.completed` field reflects only MartinLoop‑launched verifier with persisted step‑by‑step evidence.

`verification.completed` field reflects only MartinLoop‑launched verifier with persisted step‑by‑step evidence.

Source: llm_adapter@2026-06-07

Confidence: high

 —
Full changelog

Fixed

  • Verifier receipts now stay trustworthy - verification.completed now reflects the MartinLoop-launched verifier only, with persisted step-by-step evidence for launch state, exit code, timeout state, and concise command output.
  • Contradictions are surfaced instead of hidden - when adapter output says a tool or verifier failed before execution but MartinLoop’s own verifier passed, the run now keeps the pass result and records a warning/contradiction receipt instead of presenting a silent clean pass.
  • --runs-dir is consistent across the governed flow - doctor, start, preflight, run, dossier, and badge now honor the same explicit runs root without relying on MARTIN_RUNS_DIR.
  • Public help is honest again - run --help and preflight --help exit cleanly before any governed-flow checks, root-version reporting now uses the published martin-loop version, and bench is not part of the public 0.2.10 CLI surface while the public harness is being prepared.

Changed

  • CLI and MCP verification views now include verifier warnings plus step-level evidence, so operators can inspect what actually launched instead of trusting a single summary line.
  • Badge generation now uses a lightweight persisted-run probe for readiness evidence, which keeps the default command responsive even when the local runs store is large.
View diff on GitHub

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

Share on X Share on Bluesky

Track Keesan12/Martin-Loop

Get notified when new releases ship.

Sign up free

About Keesan12/Martin-Loop

All releases →

Related context

Earlier breaking changes

  • vmcp-v0.1.3 martin_status uses oneOf for selector exclusivity, latest as const.
  • vmcp-v0.1.3 maxIterations and maxTokens modeled as integers in tool schemas.
  • vmcp-v0.1.3 Tool schemas enforce additionalProperties: false on public contracts.
  • vmcp-v0.1.3 Packaged artifacts now require and ship server.json alongside package.json.

Beta — feedback welcome: [email protected]