supervise-loop

An automated supervisor-and-worker loop for Claude Code, an AI coding assistant. A supervisor sends a task to a worker, checks the result against a rubric, and uses a separate critic to provide feedback before another round.

In plain words
What is it for?
Use it for coding or other delegated tasks that can be judged with clear mechanical checks and critic feedback across repeated attempts.
Why use it?
It reduces the need for a person to inspect and restart each attempt. Work continues until the stated checks pass or a fixed limit is reached.

Plugin for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Claude Code
/plugin marketplace add wdzhwsh4067/supervise-loop
agentmods
npx agentmods add plugins/wdzhwsh4067/supervise-loop/plugin
Clone the repo
git clone --depth 1 https://github.com/wdzhwsh4067/supervise-loop

Made for: Claude Code.

Per session not measured What this adds to a session before it is invoked.
When invoked not measured Not applicable: nothing here is loaded into a session.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Security

Grade A, and why

supervise-loop scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugin/.claude-plugin/plugin.json · 28 lines

What it actually says

{
  "name": "supervise-loop",
  "version": "0.1.0",
  "description": "Autonomous supervisor/worker/critic iteration loop for Claude Code (the evaluator-optimizer pattern). Delegate a goal once; a supervisor agent grades each round against a checkable rubric — mechanical checks plus a SEPARATE critic sub-agent — and re-dispatches with feedback until the rubric passes or a hard cap is hit. No human in the loop.",
  "author": {
    "name": "王少煌",
    "email": "[email protected]"
  },
  "repository": "https://github.com/wdzhwsh4067/supervise-loop",
  "license": "MIT",
  "keywords": [
    "claude",
    "claude-code",
    "claude-code-plugin",
    "claude-skills",
    "agent",
    "multi-agent",
    "orchestration",
    "evaluator-optimizer",
    "orchestrator-workers",
    "supervisor",
    "critic",
    "agent-loop",
    "autonomous-agents"
  ],
  "homepage": "https://github.com/wdzhwsh4067/supervise-loop#readme"
}
Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 28 lines scan A 98e511c36613

Subscribe to this mod's changes

supervise-loop is a plugin published in the GitHub repository wdzhwsh4067/supervise-loop (1 stars, last pushed 2mo ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.