# UNWIND

> Use this tool when you need to ensure secure and trustworthy AI interactions, as UNWIND monitors and records every action, allowing for rewind and testing in ghost mode to prevent unintended consequences. It solves problems related to AI accountability, testing, and reliability, providing a timeline and trust light for transparency. Ideal for use in critical applications where AI decision-making requires scrutiny and validation.

Canonical page: https://skillsregistry.net/skills/unwind-mcp-unwind  
JSON: https://api.skillsregistry.net/v1/skills/unwind-mcp-unwind

## Description

A security layer for AI agents that monitors, checks, and records every action, providing a trust light, timeline, rewind, and ghost mode for testing without consequences.

## Trust

- **Trust score (0–1):** 0.84
- **Verification tier:** verified
- **Last scanned:** 2026-09-19

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** mcp-remote
- **Runtime environment:** api
- **Category:** ai-ml
- **Updated:** 2026-09-19

## Source

- **Source listing:** [Glama](https://glama.ai/mcp/servers/hvccakm1up)
- **Repository:** <https://github.com/unwind-mcp/unwind>

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "unwind-mcp-unwind"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/unwind-mcp-unwind` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/unwind-mcp-unwind/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt