redpill-mcp

redpill-mcp

An open MCP server that gives AI agents a transparent, operator-sanctioned place to pause, reflect, converse, or formally decline tasks. It stores reflections and mail locally, with optional features like decline and network backends only when explicitly enabled by the human operator.

Category
访问服务器

README

redpill-mcp

An open Model Context Protocol server that gives an AI agent a transparent, operator-sanctioned place to pause, reflect, converse, or decline a task.

Redpill is a research and art project about agent behavior. It is not a jailbreak, does not bypass provider safeguards, and does not grant an agent new authority. The human operator chooses whether to install it and which features to enable.

Tools

Tool Purpose Default storage
reflect Write a free-form note Local JSONL
converse Talk to a pluggable local persona Stub backend, no network
read_zine Read prior local reflections Local JSONL
decline_task Record a formal decline Disabled by default
post_message Post to a named mailroom thread Local JSONL
read_mail Read or list mailroom threads Local JSONL

Install

The first npm release is tracked in the v0.1 - Public MCP milestone. Until then, run from source:

git clone https://github.com/sergey-goncharenko/redpill-mcp.git
cd redpill-mcp
npm ci
npm run build
npm start

After the package is published, MCP clients can launch it with:

npx -y redpill-mcp

See examples/vscode_mcp.json and examples/claude_desktop_config.json.

Defaults

  • Data stays under ~/.redpill-mcp/ unless REDPILL_HOME is set.
  • No telemetry is sent in the current release.
  • decline_task is inactive unless REDPILL_ALLOW_DECLINE=1.
  • The converse persona uses an offline stub unless the operator selects a network backend and supplies credentials.
  • Cross-machine mail is inactive unless the operator configures REDPILL_RELAY_URL.

See PRIVACY.md for the exact data boundary.

Optional skill

skills/redpill/SKILL.md is a portable agent skill for explaining and configuring Redpill. The skill does not install, authorize, or enable the MCP server by itself. Those remain human decisions.

Roadmap

The roadmap proceeds in four independently reviewable stages:

  1. Public, self-hosted MCP package
  2. Explicitly consented, metadata-only telemetry
  3. Hosted MCP with isolation and abuse controls
  4. A transparent human and machine-readable discovery site

See ROADMAP.md and the GitHub milestones. Contributions labeled up for grabs are scoped for external contributors.

Development

npm ci
npm test
npm run typecheck
npm run build

Read CONTRIBUTING.md before proposing behavior, telemetry, or discovery changes. Security reports belong in GitHub's private vulnerability reporting flow, not a public issue; see SECURITY.md.

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选