# huangruiteng/loopx

Long-horizon agent control plane for durable, governed work across Codex, Claude Code, and other harnesses.

Repository: https://github.com/huangruiteng/loopx
Canonical: https://ross.abutalabs.com/products/loopx
Homepage: https://my.feishu.cn/wiki/CaL5wMk9ui17ngkWzeUcMlAYnZg
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agent-control-plane, ai-agents, codex, loopx, loop-engineering, agent-ops, long-running-agents, workflow-automation, agent-harness, long-horizon
Last push: 2026-08-26T21:45:53+00:00

## Health v2 (maintenance only)
Score: 80/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 6
- inputs: {"age_days": 94, "days_push": 7, "days_rel": 11, "gap_med": 1, "n_releases_24m": 42}
- flags: young
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5202, forks 464 (observed 2026-08-28T04:09:12.062882+00:00)

## What it is
LoopX is a provider-neutral, local-first control plane for long-horizon AI agents that runs on top of agent harnesses like Codex, Claude Code, and Cursor. It provides durable state, governance, recovery, and human-agent collaboration so long-running work stays reviewable and restartable across turns, tools, and agents.

## Use cases
- manage long-running AI agent goals across multiple sessions
- orchestrate Codex and Claude Code agents with durable state
- add governance and human review gates to agent workflows
- recover and resume interrupted agent work
- coordinate peer agent teams on long-horizon tasks
- track objectives, todos, and evidence for agent loops

## When to choose
- you run long-horizon agent work that spans many turns or sessions
- you need durable state, quotas, and handoffs across different agent harnesses
- you want human oversight and governance over autonomous agent loops

## When to avoid
- you only need short single-turn LLM calls without state or governance
- you want a fully autonomous agent framework rather than a control plane over existing harnesses
- you need a mature enterprise product rather than an early-stage open-source tool

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, workflow-automation, state-management, scheduling
- domain: developer-tools, large-language-models
- platform: python, cli, cross-platform
- tags: loop-engineering, agent-ops, agent-harness, long-running-agents, control-plane, codex, claude-code, local-first, ai-agents, automation

## Member repositories
- huangruiteng/loopx (main) score 80

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:12.062882+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:00:14.714576+00:00, confidence not recorded.
  - readme: https://github.com/huangruiteng/loopx (fetched 2026-08-28T04:09:12.062882+00:00, sha 45a7f1a14cec)
  - registry_pypi: https://pypi.org/pypi/loopx/json (fetched 2026-08-29T08:55:33.509213+00:00, sha f4ec76cbdf30)
- Data as of 2026-08-30T08:39:29.467469+00:00.