ULW Loop (Evidence-Bound Long-Run Orchestrator)

A durable goal-execution skill that decomposes work into atomic steps, requires real observable evidence for every success criterion, and delegates to Codex subagents.

AutomationAdvanced★ 67,531⑂ 5,507AI score 7/10Last updated: Aug 9, 2026

What it does

Registers goals up front and mirrors every atomic action into a live checklist with exactly one step in_progress.
Rejects "tests passed" as proof of done — each success criterion needs evidence from a real surface (terminal/TUI, HTTP, browser, CLI stdout, DB diff), bound to the current git tree hash so it goes stale when tracked content changes.
Enforces a strict subagent delegation contract: every spawn_agent message starts with TASK: and names DELIVERABLE, SCOPE, VERIFY; long-running plan/reviewer agents run in the background with exponential wait back-off.
Defines recovery after context compaction: re-read brief, goals, ledger and ulw-loop status --json before resuming, never re-plan from scratch.
Gives an explicit decision rule for team mode (overlapping scopes + true parallel speedup), one git worktree per member, and per-unit merges with the lead owning conflicts.
Maps orchestration intents to both MultiAgentV2 and multi_agent_v1.* tool surfaces.

Who it's for

Engineers running Codex / oh-my-openagent with multi-agent tooling.
Anyone doing multi-session refactors or migrations where context loss is routine.
Teams that want observable-evidence gates instead of green-CI optimism.

Examples

"Run ulw-loop on the payments migration" → goals registered, per-step checklist, real run logs recorded as evidence per step.
Large PR review → reviewer agent spawned in the background while the root agent continues independent work, tracking WORKING: phases.
Session resumed after compaction → read ledger plus status --json, then continue from the last in_progress step.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/code-yeongyu/oh-my-openagent/HEAD/packages/omo-codex/plugin/components/ulw-loop/skills/ulw-loop/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the packages/omo-codex/plugin/components/ulw-loop/skills/ulw-loop folder from the GitHub repo code-yeongyu/oh-my-openagent into my ~/.claude/skills/ulw-loop/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/code-yeongyu/oh-my-openagent.git /tmp/oh-my-openagent && mkdir -p ~/.claude/skills && cp -r /tmp/oh-my-openagent/packages/omo-codex/plugin/components/ulw-loop/skills/ulw-loop ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Open a terminal.
Clone the repo: git clone https://github.com/code-yeongyu/oh-my-openagent.git
Copy the skill folder: mkdir -p ~/.claude/skills && cp -r oh-my-openagent/packages/omo-codex/plugin/components/ulw-loop/skills/ulw-loop ~/.claude/skills/
Verify ~/.claude/skills/ulw-loop/references/full-workflow.md came along — the skill is incomplete without it.
Make sure the omo-agent-toolkit CLI is installed (try omo-agent-toolkit ulw-loop status --json).
Restart Claude Code and ask for "ulw-loop" or "ulw" on a task to trigger the skill.

View source on GitHub ↗License: NOASSERTION