Performance Goal Workflow

SkillDev tools

The performance-goal skill runs an evaluator-gated performance optimization workflow over Codex goal mode with durable OMX artifacts and safe goal handoffs. It is a skill for an AI agent that sets up a structured tuning loop under .omx goals, requiring an evaluator command and pass/fail criteria before any changes begin. The agent makes small reversible code changes, runs benchmarks and regression tests, records pass/fail/blocked checkpoints, and completes the goal only after a verified passing result.

Available today. Use it from your connected AI after setup.

Have an evaluator command and a pass/fail contract ready before starting.

Then ask your AI: use the Performance Goal Workflow skill

What your AI can do with it

  • Create a performance workflow with omx performance-goal create
  • Emit a Codex goal handoff with omx performance-goal start
  • Record pass, fail, or blocked evidence with omx performance-goal checkpoint
  • Complete a goal only after a passing checkpoint with omx performance-goal complete
  • Store durable workflow state under .omx/goals/performance/<slug>/

Getting started

  1. Have an evaluator command and a pass/fail contract ready before starting.
  2. Run omx performance-goal create with the objective, evaluator command, evaluator contract, and a slug.
  3. Run omx performance-goal start and follow the handoff to set up the active goal.
  4. Make small reversible code changes, then run the evaluator and related regression tests.
  5. Record each result with omx performance-goal checkpoint and complete only after a passing checkpoint.

What this skill tells your AI

The instructions your AI receives, as published by yeachan-heo/oh-my-codex in skills/performance-goal/SKILL.md and read by ahel’s review.

Use this skill when a user asks OMX to optimize performance and wants a goal-oriented loop rather than a one-off review.

Contract

  • OMX owns durable workflow state under .omx/goals/performance/<slug>/.
  • Codex goal mode owns only the active-thread focus/accounting primitive.
  • Shell commands do not mutate hidden Codex goal state. They write artifacts and emit model-facing handoff text.
  • No optimization work may start until an evaluator command and pass/fail contract exist.
  • Do not call update_goal({status: "complete"}) until the evaluator has a passing checkpoint and a completion audit proves the objective is done; then call get_goal again and pass that fresh snapshot to omx performance-goal complete --codex-goal-json.

CLI

Create the workflow and evaluator contract:

omx performance-goal create \
  --objective "Reduce CLI startup latency by 20%" \
  --evaluator-command "npm run perf:startup" \
  --evaluator-contract "PASS when p95 latency improves by 20% and regression tests pass" \
  --slug startup-latency

Emit the Codex goal handoff:

omx performance-goal start --slug startup-latency

Record evaluator evidence:

omx performance-goal checkpoint --slug startup-latency --status pass --evidence "benchmark + tests passed"
omx performance-goal checkpoint --slug startup-latency --status fail --evidence "benchmark regressed"
omx performance-goal checkpoint --slug startup-latency --status blocked --evidence "missing fixture"

Complete only after a passing checkpoint:

omx performance-goal complete --slug startup-latency --evidence "final evaluator evidence" --codex-goal-json <get_goal-json-or-path>

Agent Loop

  1. Run omx performance-goal create if no workflow exists.
  2. Run omx performance-goal start and follow the handoff:
    • call get_goal;
    • call create_goal only when no active goal exists and the objective is explicit;
    • work only against the evaluator contract;
    • after evaluator pass and completion audit, call update_goal({status: "complete"}), call get_goal again, and pass that snapshot to omx performance-goal complete --codex-goal-json;
    • after omx performance-goal complete succeeds, run /goal clear in the Codex UI before starting another goal in this same thread/session; OMX prints this terminal cleanup step but does not invoke hidden clear routes;
  3. Optimize in small reversible patches.
  4. Run the evaluator and related regression tests.
  5. Record each pass/fail/blocker with checkpoint.
  6. Complete only when the pass artifact exists and no required work remains.

Completion Gate

A performance goal is incomplete unless .omx/goals/performance/<slug>/state.json contains a lastValidation.status of pass and omx performance-goal complete receives a matching complete Codex get_goal snapshot via --codex-goal-json. Passing ordinary tests alone is not sufficient unless they are the declared evaluator contract.

Lifecycle: create_goal starts the Codex thread goal, update_goal({status: "complete"}) marks terminal success after the evaluator and audit pass, and /goal clear removes the completed thread goal when another same-thread goal is needed. OMX shell commands and hooks reconcile snapshots and print the cleanup instruction; they must not mutate hidden Codex goal state.

Signals

GitHub stars
33k
Forks
3k
Last commit
Oct 2026

Questions

What is performance-goal?
It is a skill that runs an evaluator-gated performance optimization workflow over Codex goal mode with durable OMX artifacts and safe goal handoffs.
When should I use this skill?
Use it when a user asks OMX to optimize performance and wants a goal-oriented loop rather than a one-off review.
Advanced
Item type
skill
Key
performance-goal
Source
github.com/yeachan-heo/oh-my-codex