Clew

mcp
Security Audit
Pass
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 18 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool acts as an AI-assisted workspace that generates interactive learning roadmaps and dependency graphs from your study goals or notes.

Security Assessment
Overall risk: Low. The project requires you to supply your own AI provider API keys (such as OpenAI or Gemini) via a local environment file, meaning there are no hardcoded secrets. The light code audit scanned 12 files and found no dangerous patterns, no suspicious permission requests, and no evidence of unauthorized shell command execution. The only expected network requests are the standard API calls to your chosen AI provider, and it securely defaults to a local SQLite database.

Quality Assessment
The project is actively maintained with a recent push occurring today. It uses the permissive MIT license and has accumulated 18 GitHub stars, indicating a small but present level of community validation. The repository is well-documented and provides clear, straightforward setup instructions for local deployment.

Verdict
Safe to use.
SUMMARY

Clear path from where you are to what you want to learn. AI agent turns goals and topic dumps into graph with visible dependencies (generative roadmap.sh)

README.md

Clew

MIT CI Generative Roadmap Agentic Workspace Knowledge Graph Obsidian integration

Generate a learning map. Click any topic. Follow the thread back to its foundations.

Clew walkthrough

Clew is a local, AI-assisted workspace for studying through dependency graphs. Start from a goal, a rough topic dump, or an Obsidian vault; Clew turns it into a map of topics, prerequisites, resources, artifacts, and progress.

The graph is the workspace. AI can draft, expand, audit, and reshape it, but changes stay visible, reviewable, and reversible.

Try the hosted version at clew.my, or run this repo locally when you want provider control, local state, Obsidian import/export, and MCP context.

Quick Look

  • Core idea: a visible thread through hard subjects
  • Main move: click a topic and see the path that leads to it
  • Graph generation: build a map from a goal, notes, topic list, or Obsidian vault
  • AI boundary: AI proposes structure; you review and apply changes
  • Local edition: SQLite, provider keys, Gemini/OpenAI support, import/export, MCP

Quick Start

git clone https://github.com/miuuyy/Clew.git
cd Clew
cp .env.example .env
./scripts/dev.sh

Set one provider key in .env:

KG_GEMINI_API_KEY=...
# or
KG_OPENAI_API_KEY=...

Then open:

  • frontend: http://127.0.0.1:5178
  • backend: http://127.0.0.1:8787

Why Clew Exists

Learning a big subject is not just about collecting resources. The hard part is structure.

A chat answer can tell you what to read. Tools like roadmap.sh show a common route. But when your goal is specific, you need to see what actually unlocks what: which foundations matter now, which topics can wait, where you are blocked, and how far you are from the thing you want to build.

Clew makes that structure visible.

Instead of manually arranging a huge roadmap, you let AI draft the graph. Then you study through it: click topics, inspect prerequisite paths, attach resources, pass quizzes, mark progress, and keep the whole learning process tied to the map.

Features

  • Click-to-path navigation: select any topic and reveal the prerequisite chain behind it.
  • Generative roadmaps: create a dependency graph from a goal, topic dump, notes, or vault.
  • Graph-first workspace: topics, dependencies, zones, resources, artifacts, layout, and progress live on one surface.
  • Reviewable AI changes: ingest, expand, audit, reshape, apply, and roll back through snapshots.
  • Study loop: topic sessions, assistant help, inline quizzes, closure quizzes, and manual completion when strict gating is disabled.
  • Obsidian bridge: import a vault into Clew or export a graph back into an Obsidian-ready folder.
  • Obsidian-to-Clew import skill: packaged for Claude Code and Codex to audit a vault, flag blockers, and shape it into a valid Clew package.
  • MCP context bridge: let Claude, Cursor, or another MCP client read your Clew graphs and progress without copy-paste.
  • Local control: SQLite workspace, provider keys, Gemini/OpenAI support, OpenAI-compatible endpoint option, memory/persona/thinking settings.

Example Use Cases

  • Turn "I want to build a machine learning project" into the math, programming, and ML path that actually matters.
  • Click a hard topic and see the foundations you are missing.
  • Build a Python, cybersecurity, systems, math, or exam roadmap with visible prerequisites.
  • Convert a messy topic dump into a graph you can actually study through.
  • Import an Obsidian vault and check whether your notes form a usable learning structure.
  • Ask an external assistant about your current learning path through MCP.
  • Export a finished path back to Obsidian as readable notes.

Visuals

Clew workspace overview

Clew topic view Clew proposal flow Clew path highlight

Docs

Repository Map

Path Role
frontend/ React workspace UI, graph canvas, themes, settings, dialogs, debug surfaces
backend/ FastAPI app, repository, domain model, provider layer, planner, MCP server, tests
contracts/ JSON contracts and transport surfaces used by graph mutation flows
docs/ Engineering docs, ADRs, release notes, and site FAQ source
scripts/ Local development helpers such as boot, stop, and reset
.claude/skills/obsidian-to-clew-import/ Claude Code skill for shaping an Obsidian vault into a Clew import package
.agents/skills/obsidian-to-clew-import/ Codex-compatible copy of the Obsidian-to-Clew import skill

Development Checks

cd frontend && npm run typecheck && npm run build
PYTHONPATH=backend ./.venv/bin/python -m unittest discover -s backend/tests -v

Useful helpers:

./scripts/dev.sh
./scripts/stop_dev.sh
./scripts/reset_db.sh

Open Source

Contact

Reviews (0)

No results found