codexmon v0.12.0

Added

• codexmon models --agent codex|claude [--json] lists current model choices
  and marks the default. New choices include GPT-6.1 Sol, Claude Opus 5.5,
  and Claude Sonnet 5.5. Native model overrides remain unrestricted.
  Text output contains only the table; the informational note goes to stderr.
• The README and bundled skill list the current model choices. GPT-6 Astra
  and Claude Fable 5.1 remain the defaults for demanding work.

Changed

• models now belongs to codexmon. Cursor's native models command still
  runs through the monitor. Use codexmon run --agent cursor -- models to
  reach Cursor's native command directly.

codexmon v0.11.0

Changelog

• 42ffb2c: feat: default Claude Code to Fable 5.1 and release v0.11.0 (@tigercosmos)

codexmon v0.10.0

Codex exec runs now default to GPT-6 Astra (gpt-6-astra) with high reasoning effort. Explicit model, profile, and effort overrides remain supported.

Updated the README, bundled agent skill, version metadata, and regression tests.

Validation:

• Live Codex CLI 0.153.4 calls with default high and explicit ultra both completed with exit code 0 and the exact expected output.
• make all, make lint, and make race passed locally.

Full changelog

codexmon v0.9.0

Codex runs can now select every effort level in the Codex model catalog.

Highlights

• Add --effort to monitored run, start, and review commands.
• Support low, medium, high, xhigh, max, and ultra.
• Map the option to Codex's model_reasoning_effort setting.
• Reject invalid values, duplicate native settings, and unsupported agents.

Example

codexmon review --agent codex --effort ultra --uncommitted

Verification

Real end-to-end smoke tests passed with Codex 0.151.0, Claude Code 2.1.252,
and Cursor Agent 2026.08.11. Linux and macOS CI also passed.

Full changelog: v0.8.0...v0.9.0

codexmon v0.8.0

Changelog

• d122dc5: fix: harden process lifecycle, agent fallback, and retention (@tigercosmos)

codexmon v0.7.0

Changelog

codexmon v0.6.0

Changelog

• 7181ef0: feat: fall back across agents when no backend is specified (@tigercosmos)

codexmon v0.5.0

Changelog

• 2508a2d: feat: default codex exec to gpt-5.6-sol at high reasoning effort (@tigercosmos)

codexmon v0.4.0

Changelog

• e6ff468: feat: job retention, adaptive wait, and one-call JSON result (@tigercosmos)

codexmon v0.3.0

Changelog

• cbae3c1: feat: default Cursor agent reviews to Composer 2.5 (@tigercosmos)