Agent Loop Skill
- 5 repo stars
- Updated June 30, 2026
- dososo/agent-loop-skill
A workflow skill for long-running Codex, Claude Code and Cursor tasks: contracts, disk-backed state, role separation, verification traces, restarts, subjective scoring and cleanup.
About
Agent-loop-skill is a workflow scaffold for running coding agents on long, multi-step tasks reliably. It gives a solo developer contracts, disk-backed state, role separation between planner and worker, verification traces, automatic restarts, subjective scoring of results and cleanup, so an agent can grind through a long job without losing its place. It targets Codex, Claude Code and Cursor, and you reach for it when a single agent run is too long or fragile to trust in one pass.
- Contracts and role separation for long-running agent tasks
- Disk-backed state so work survives restarts
- Verification traces and subjective scoring of runs
- Works across Codex, Claude Code and Cursor
Agent Loop Skill by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add dososo/agent-loop-skillAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| repo stars | ★ 5 |
|---|---|
| Last updated | June 30, 2026 |
| Repository | dososo/agent-loop-skill ↗ |
What it does
A workflow skill for long-running Codex, Claude Code and Cursor tasks: contracts, disk-backed state, role separation, verification traces, restarts, subjective scoring and cleanup.
Who is it for?
Solo devs running agents on long multi-step jobs that need restarts
Skip if: Quick one-shot prompts that finish in a single agent turn
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
