#7474th of 149 in Skills, plugins and rules
loopx
Local control plane that keeps goals and evidence for long-running coding agents
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 6.3k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- Python
11 on 3 Jun 20266,316 on 14 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
LoopX installs as a skill for Codex, Claude Code and DeepSeek Harness. It stores goals, Todos, decisions and evidence outside the chat session, so an agent can continue across runs through a heartbeat or loop driver and ask the owner when a gate is hit. It includes a local browser/PWA dashboard (`loopx dashboard`) and a CLI (`loopx status`, `loopx doctor`).
Who it is for: Engineers running multi-day agent loops in Codex or Claude Code
Strengths
- Works with Codex, Claude Code and DeepSeek Harness without replacing their runtime
- Goal state, decisions and evidence persist across sessions and restarts
- Local-first dashboard shows what is running, waiting on you, or scheduled
- Publishes LHTB results: 0.4948 mean reward vs 0.4218 for plain Codex on 46 tasks
Weaknesses
- LHTB uses one trial per task and unequal budgets; not an isolated causal estimate
- Strict pass rate (7/46) matches plain Codex; the gain is mostly partial credit
- Needs Python 3.11+ and Node.js 22.22.3+ installed
- Desktop App is ad-hoc signed, not notarized; Windows preview updates are manual
What it needs
- Needs Python 3.11+, Node.js 22.22.3+, Codex, Claude Code or DeepSeek Harness
- Models: Codex, Claude Code, DeepSeek Harness, direct-model sessions
Also in Skills, plugins and rules
See all 149| Rank | Project | Score |
|---|---|---|
| 1 | cavemanCuts token use in coding agents with terse replies and compressed inputs | 95 out of 100 |
| 2 | skillsAgent skills for planning, TDD, debugging and code review | 94 out of 100 |
| 3 | eccAgents, skills, hooks and rules bundle for Claude Code, Codex and other harnesses | 94 out of 100 |
| #4 | impeccableDesign skill with 24 commands and a detector for AI-generated frontends | 94 out of 100 |
| #5 | ponytailSkill prompt that makes coding agents write less code and add tests | 92 out of 100 |