Anneal
io.github.imsovikde由 GitHub 账号 imsovikde 发布v1.2.6更新于 Oct 24, 2026
Zero-friction personal operating system: habits, tasks, and FSRS-5 study retention via remote MCP.
概览
Anneal 的远程 MCP 端点,让助手管理个人习惯、任务以及基于 FSRS-5 的间隔复习学习数据。
- 可以做什么
- 创建并记录习惯,查看坚持度、连续天数和行为趋势
- 整理任务与截止日期、安排优先级并记录完成情况
- 创建卡组与卡片、获取到期卡片、评分复习并按 FSRS-5 安排间隔
- 生成学习大纲、管理文件夹并在卡组间移动卡片
- 查看账户资料、导出数据、查看审计日志与存储配额
- 列出并查看服务器自身的工具与能力
- 适用场景
- 适合已在 Anneal 中记录习惯、任务或学习材料,并希望助手在对话中读取和更新这些数据的场景,例如每日打卡回顾、连续记录检查、抽认卡测验或为某科目制定学习计划。
- 运行要求
- 远程 streamable HTTP 端点,无需本地包或运行时。需要 Anneal 账户,并通过 OAuth 2.1 与 PKCE 授权同意;桌面客户端可在 Authorization 头中填入 Bearer API 密钥。需能访问该托管端点。
安装
在 SourceWeft 中
- 打开 控制台中的 Anneal,将其添加到工作区。
- 连接 所需的 凭据。它们会被私密存储,绝不会在公开页面上展示。
- 为需要使用其工具的对话启用该服务。
Web executable,通过 Streamable HTTP。 远程服务在工作区中配置后即可从网页运行时运行。
交给智能体安装
把这段提示词粘贴到 Claude Code、Codex、Cursor 等智能体,或 SourceWeft 对话中。智能体会读取该服务器的安装指南,先向你展示发布者、端点和登录方式,经你同意后再把它添加到你的客户端。密钥不会写进任何命令。
阅读 https://sourceweft.com/mcp/io.github.imsovikde%2Fanneal/install.md,并按其中说明把这个 MCP 服务器接入我的客户端。修改任何配置之前,先告诉我它的发布者、端点和登录方式,等我确认。不要把 API 密钥写进命令或我会提交的文件;问我密钥应从哪里获取。其他 MCP 客户端
把它添加到你客户端的 mcpServers 配置中。 客户端会在首次连接时完成 OAuth 登录。
{
"mcpServers": {
"anneal": {
"type": "http",
"url": "https://anneal-os.vercel.app/api/mcp"
}
}
}README
[Anneal MCP Official Vector Mark]
Anneal MCP Server
Official Model Context Protocol Remote Server for the Anneal Personal Operating System
Connect Claude, ChatGPT, Gemini, and Cursor directly to your unified personal OS — habits, task lifecycles, and FSRS-5 spaced repetition retention in a single append-only ledger.
[CI Manifest Check] [Protocol Specification] [Transport Type] [Security Standard] [Gemini Compatible] [License]
Instant 1-Click AI Client Integrations
[Add to Claude] [Add to ChatGPT] [Add to Gemini] [Install with Smithery]
Live Web Application · Server Endpoint · Architecture Guide · Submit Feedback
What is Anneal MCP?
The Anneal Model Context Protocol (MCP) Server is an enterprise-grade remote integration endpoint that allows frontier AI models (Anthropic Claude, OpenAI ChatGPT, Google Gemini, and Cursor) to read, record, and optimize your personal data stream.
Unlike fragmented tools that isolate your tasks in one app, your habits in another, and your study revision in a third, Anneal unifies all three disciplines in a single, append-only event ledger. Through this server, AI agents act directly on your behalf:
- Habits Engine: Query adherence, log completions, calculate streaks, and project behavioral velocity.
- Task Lifecycle: Organize deadlines, prioritize discrete deliverables, and record completed tasks with zero modal friction.
- Cognitive Retention Engine: Generate comprehensive academic syllabi, create flashcard decks, and schedule optimal review intervals powered by FSRS-5 (Free Spaced Repetition Scheduler).
Competitive Architecture Benchmark
Why frontier AI agents perform better with Anneal MCP compared to legacy productivity integrations:
Technical Specifications
Client Setup Guides
1. Anthropic Claude (Web & Desktop)
A. Claude.ai Web Interface (1-Click)
- Click the 1-Click Claude Connect deep-link.
- Claude.ai will open the Add custom connector modal with name
Annealand the production URL pre-filled. - Click Add, grant consent, and start using Anneal tools in any conversation.
B. Claude Desktop (claude_desktop_config.json)
Add the following entry to your configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS or %APPDATA%\Claude\claude_desktop_config.json on Windows):
2. OpenAI ChatGPT
- Navigate to ChatGPT Settings → Connected Apps / Plugins.
- Click Add Remote Tool / Plugin.
- Input server endpoint:
- Authenticate via standard OAuth 2.1 consent screen.
3. Google Gemini & Gemini Spark
Install directly via the Gemini CLI:
Gemini Spark Note: Anneal includes automatic Protobuf payload sanitization to prevent gRPC 13 schema exceptions.
4. Cursor & Windsurf IDEs
Install using the Smithery package runner:
5. Cline / Roo Code / Continue.dev
Add the remote HTTP endpoint to your IDE extension settings:
Tool Catalog & Modules (102 Tools)
Anneal exposes 102 typed tools organized across 5 domain modules:
AI Agent Prompt Cookbook
Copy and paste these verified interaction recipes into Claude, ChatGPT, or Gemini:
Scenario 1: Habit Tracking & Reflection
"Inspect my Anneal habits for the past 14 days. Show my current streaks, identify any habit at risk of lapse, and log a 45-minute meditation session for today."
Scenario 2: FSRS-5 Spaced Repetition Review
"Check Anneal for all flashcards due for review today in my 'Machine Learning' deck. Test me one card at a time, grade my answers, and record my recall rating using FSRS-5."
Scenario 3: Automated Syllabus Generation
"I need to master Linear Algebra in 6 weeks. Generate a structured syllabus in Anneal with daily study targets, create flashcard decks for each chapter, and schedule initial retention reviews."
Frequently Asked Questions (FAQ)
How does Anneal protect user privacy during AI interactions?
Anneal implements zero-trust authorization. All requests are authenticated via OAuth 2.1 with PKCE. Tools only operate within the authenticated user's ActorContext. No database credentials or cross-tenant records are ever accessible.
Does Anneal delete historical logs when habits or tasks are removed?
No. Anneal uses an immutable append-only event ledger. Deletions create tombstone markers, and updates record forward pointers (supersedesId). Historical records remain intact for longitudinal data visualization.
What spaced repetition model does Anneal use?
Anneal uses FSRS-5 (Free Spaced Repetition Scheduler), an advanced 9-parameter memory algorithm that models memory stability and retrievability based on cognitive decay curves, outperforming legacy SM-2 models by over 18%.
Registry Indexing & Discovery
Anneal MCP is published and indexed across primary registries:
- Model Context Protocol Registry:
modelcontextprotocol/registryviaserver.json - Smithery:
smithery.aiviasmithery.yaml - Google Gemini Extensions: via
gemini-extension.json
License
This MCP distribution repository and its client manifests are licensed under the MIT License.
来源:README.md,提交 353c1b6
工具
0版本历史
1- v1.2.6最新Sep 16, 2026
评分
已满足适用检查项的 85%
该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。
- 近 30 天内可连通必需25 分
- 已声明认证方式必需10 分
- 已列出工具15 分
请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。
- 描述必需15 分
- 图标10 分
- 文档15 分
- 许可证10 分