Memory and knowledge · MCP server
ThoughtDAG: MCP server, setup and security
ThoughtDAG is an open-source MCP server published on GitHub that lets an AI agent find past agent conversations by file or exact phrase. Its repository has 693 stars and its last commit is from October 24, 2026 and it is licensed under MIT.
| Adoption (stars) | 28/40 | |
|---|---|---|
| Maintenance | 30/30 | |
| Framework (licence, registry, official) | 20/30 |
What ThoughtDAG does
According to its repository: “Search local AI-agent conversation history by file or exact phrase and recall original source turns through four read-only tools.”
The Model Context Protocol (MCP) is a standard plug between an AI agent and an outside service: once the server is connected, its tools show up in the agent, which decides when to call them. ThoughtDAG is listed under Memory and knowledge, a category of 250 servers in the directory, where it ranks #21 by stars.
Topics declared by the repository: agentic-search, ai-tools, context-engineering, dag, infinite-canvas, knowledge-graph.
Installing and connecting ThoughtDAG
Distribution modes declared in the official MCP registry:
- npm package (Node.js, usually started with
npx)
It works with any MCP-compatible client (Claude Desktop, Claude Code, Cursor, VS Code, Codex and most recent agents). You add one entry to the client's MCP configuration: server name, start command or URL, environment variables. Project language: TypeScript. Listed systems: macOS, Windows, Linux.
Security: what to check
Connecting an MCP server means installing software that talks straight to your agent: it runs with your rights and the model trusts its tool descriptions. The risk therefore depends on what it can touch and how far its code can be trusted.
None of the risk signals we track (system access, secrets, missing licence, abandonment) applies to ThoughtDAG. That does not replace reading the code.
Good practice: grant only the rights needed (least privilege), require human approval for irreversible actions, and run the AI agent security check.
Project activity and maintenance
Created on February 17, 2026, the chenxiachan/thoughtdag repository has 693 stars and 68 forks, with 2 open issues. It has more stars than 90% of the servers in the directory. It received changes this month. Version published in the registry: 0.2.2 (September 27, 2026).
Stars measure attention, not quality: a recent, heavily promoted project can collect them fast, while a useful but niche server will have few.
Verdict
ThoughtDAG is one of the most solid servers in its category: followed, maintained and well framed.
Strengths
- Active project: recent commit
- Widely followed: 693 stars
- Listed in the official MCP registry
- Permissive licence (MIT)
Alternatives
- Context763.6K ★
- Basic memory4.3K ★
- Hindsight53.4K ★
- Cognee32.7K ★
- Skill Seekers15.2K ★
- Honcho7.8K ★
Learn more
Frequently asked questions
How do I install ThoughtDAG?
ThoughtDAG is distributed as npm package (Node.js, usually started with npx). Add it to your client's MCP configuration (Claude Desktop, Cursor, VS Code…) following the README of chenxiachan/thoughtdag.
Is ThoughtDAG free?
The code is published under the MIT licence and can be used free of charge.
Is ThoughtDAG safe?
No risk signal found (system access, secrets, licence, abandonment). As with any MCP server, keep rights minimal and read the code before sensitive use.
Is it an official server?
Nothing says so: it is a project published by chenxiachan on GitHub. Check who maintains it before giving it access.
What are the alternatives to ThoughtDAG?
In the same category: Context7, Basic memory, Hindsight, and Cognee. Context7 scores higher on our grid.
Related articles
- Build your own MCP server in 30 minutesOctober 24, 2026
- Automate your inbox with an AI agentOctober 18, 2026
- MCP, A2A, and tool calling: agent protocols explainedOctober 17, 2026
- Build an AI agent with n8n, without codingOctober 16, 2026
Neutrality. No affiliate links. No vendor pays to be reviewed, rated or ranked. Reviews are directed by François Chollet, neutral and factual, and every data point is dated and sourced. How we score