Skip to content

This release includes 1 security fix for security teams reviewing exposed deployments.

✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →
This release patches 1 known CVE

Affected surfaces

auth breaking_upgrade

Summary

AI summary

Codex project‑scope trust bootstrap now keeps AIRG‑managed tool approvals local to the trusted project .codex layer, enhancing security.

Full changelog

[2.3.1] - 2026-05-06

Changed

  • Promoted the 2.3.1 development train to release and aligned package, status, hook, and documentation version markers.
  • Telemetry architecture moved to an hourly scheduler that runs generator/uploader workers in parallel.
  • Telemetry state now tracks telemetry.last_payload_generated_date and telemetry.last_payload_uploaded_at in policy.
  • Codex integration is now scope-aware across MCP apply/remove, hardening, and posture detection.
  • Codex hardening now uses dedicated scoped rules/airg.rules policy-mirror files instead of managing Codex default.rules.

Added

  • Telemetry outbox persistence in the runtime data directory (<state_dir>/telemetry/telemetry-YYYY-MM-DD.json).
  • Telemetry service status and restart API endpoints:
    • GET /telemetry/service-status
    • POST /telemetry/service-restart
  • Advanced Policy telemetry UI now includes:
    • service status modal (status + last run for generator/uploader)
    • warning banner for stale generator/failing uploader
    • restart button.
  • Codex project-scope trust bootstrap in ~/.codex/config.toml via [projects."<workspace>"].trust_level = "trusted" with explicit GUI confirmation.
  • Codex project-only AIRG tool approval entries in <workspace>/.codex/config.toml for [mcp_servers.ai-runtime-guard.tools.*].

Fixed

  • Telemetry no longer depends on UTC day-rollover wake logic only; hourly scheduler runs continue while backend is active and stand down when no action is needed.
  • Upload retries now persist naturally through outbox files instead of dropping failed sends.
  • Fixed write_file/edit_file tool crashes caused by NameError: name 'lower' is not defined in check_path_policy extension checks.
  • Reports -> Log table now enforces fixed-width column truncation for long event/rule/command/path values so oversized text no longer expands row/cell boundaries.
  • Reports -> Log now exposes consistent Show more actions for trimmed Event, Matched Rule, and Command / Path values via modal expansion.
  • Codex text rewrite/apply flows are now idempotent across scoped config.toml, AGENTS.md, and rules/airg.rules.
  • Settings -> Agents no longer shows false Pending Changes for hardening state when posture already matches the applied baseline.

Security

  • Codex project scope now keeps AIRG-managed tool approvals local to the trusted project .codex layer instead of writing them to user ~/.codex/config.toml.

Tests

  • Added regression coverage for:
    • telemetry generator/uploader scheduler behavior
    • blocked-extension write/edit regression after path-policy fix
    • Codex project-trust apply/remove behavior
    • Codex project-scope hardening targets and idempotency
    • Codex project-scope posture detection
    • Codex project-only tool approval sections.

Security Fixes

  • Codex project scope now confines AIRG‑managed tool approvals to the trusted project `.codex/config.toml` layer instead of writing them to user `~/.codex/config.toml`, preventing cross‑project approval leakage.

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 jimmyracheta/AI-Runtime-Guard

Get notified when new releases ship.

Sign up free

About jimmyracheta/AI-Runtime-Guard

Runtime policy enforcement for AI agents - prevents accidental damage to your systems, unauthorized agent access and automates backup-before-write for any touched files.

All releases →

Related context

Beta — feedback welcome: [email protected]