Skip to content
Back to skills

Clade Control

ASecurity

Start, stop, or manage Clade autonomous coding loops remotely

  • 10 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added June 11, 2026
ai-agentsgoapi

Works with

  • api

Security analysis

A100/100

Scanned June 11, 2026

npx -y skills add shenxingy/Clade --skill clade-control --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Clade Control?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Clade Control
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/shenxingy-clade-control/badge)](https://www.skillsdirectory.com/skills/shenxingy-clade-control)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: clade-control
version: 1.0.0
description: Start, stop, or manage Clade autonomous coding loops remotely
author: alexshen
tags: [claude-code, orchestrator, automation, devtools, coding-automation]
requires:
  env:
    - CLADE_BASE_URL
    - CLADE_API_KEY
---

# Clade — Control

Start, stop, or manage autonomous coding loops on a remote machine.

## When to use

User wants to:
- Start a new coding loop with a goal
- Stop a running loop
- Clear blockers so a loop can resume

## API

**Endpoint:** `POST {CLADE_BASE_URL}/control`

**Headers:**
```
Authorization: Bearer {CLADE_API_KEY}
Content-Type: application/json
```

### Action: start

Start a new autonomous coding loop with a goal.

**Request:**
```json
{
  "project": "/home/user/myproject",
  "action": "start",
  "goal": "Fix all failing tests and ensure 100% pass rate",
  "max_iter": 5,
  "max_workers": 4,
  "model": "sonnet"
}
```

**Translating user intent to parameters:**
- "run 3 times" / "跑3次" → `max_iter: 3`
- "use opus" / "用opus" → `model: "opus"`
- "2 workers" / "2个worker" → `max_workers: 2`
- "quick run" → `max_iter: 3, model: "haiku"`
- "thorough run" → `max_iter: 10, model: "opus"`
- If user doesn't specify: default to `max_iter: 10, max_workers: 4, model: "sonnet"`

**The goal can be:**
- A natural language description: `"Fix the login page CSS on mobile"`
- A file path on the server: `"/home/user/myproject/goals/feature-x.md"`

**Response:**
```json
{
  "ok": true,
  "action": "start",
  "pid": 12345,
  "goal": "/home/user/myproject/.claude/monitor-goal.md",
  "max_iter": 5
}
```

### Action: stop

Stop the currently running loop gracefully.

**Request:**
```json
{
  "project": "/home/user/myproject",
  "action": "stop"
}
```

**Response:**
```json
{"ok": true, "action": "stop", "sentinel": "/home/user/myproject/.claude/stop-start"}
```

### Action: clear-blockers

Clear blockers so the loop can be restarted.

**Request:**
```json
{
  "project": "/home/user/myproject",
  "action": "clear-blockers"
}
```

## How to format the response

### On start:
```
🚀 Loop started!
📋 Goal: {goal summary, first 80 chars}
⚙️ Config: {max_iter} iterations, {max_workers} workers, {model}
🔢 PID: {pid}

Use "status" to check progress.
```

### On stop:
```
🛑 Stop signal sent. The loop will finish its current task and exit gracefully.
```

### On clear-blockers:
```
🧹 Blockers cleared. You can now restart the loop.
```

## Error handling

- **Connection refused**: "Monitor is not running."
- **401 Unauthorized**: "Check your CLADE_API_KEY."
- **400 "goal is required"**: Ask the user what they want the AI to work on.
- **500 "start.sh not found"**: "Clade is not installed on the server. Run ./install.sh first."

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.