🐍 Compython

AI chat assistant for Obsidian with vault awareness, semantic search, one-click edits, and native Python execution.

Compython is a fork of Smart Composer that adds the ability to write and run Python code directly inside Obsidian — both manually by the user and autonomously by the AI model.

---

✨ What's new in Compython

Native Python Execution

Every Python code block in the chat gets a "Run Python" button. Click it and the output appears inline — no terminal, no switching apps, no copy-paste.

• Works with any AI model — local (Gemma, Qwen, LLaMA) or cloud (Claude, GPT, Gemini)
• No tool calling required — based on markdown code block detection, so every model works
• Output appears inline with stdout, stderr, and exit code
• The vault path is automatically available as the VAULT_PATH environment variable inside scripts

---

🚀 Features (inherited from Smart Composer)

• Contextual Chat — reference vault files with @filename syntax
• Apply Edit — AI suggests edits to your notes with one-click apply
• Vault Search (RAG) — semantic search across your entire vault
• MCP Support — connect to external tools via Model Context Protocol
• Local Model Support — Ollama, LM Studio, and any OpenAI-compatible provider
• Custom System Prompts and Prompt Templates
• Multiple AI providers: Anthropic, OpenAI, Gemini, Groq, DeepSeek, OpenRouter, and more

---

📦 Installation

Manual installation

1. Download the latest release: main.js, manifest.json, styles.css
2. Create the folder: YOUR_VAULT/.obsidian/plugins/compython/
3. Copy the 3 files into that folder
4. In Obsidian: Settings → Community plugins → reload → enable Compython

⚠️ If you have Smart Composer installed, disable or uninstall it first — they conflict since they register the same view type.

---

⚙️ Setting up Python execution

1. Go to Settings → Compython → Python Executor
2. Toggle Enable Python execution ON
3. Set the Python path:
   • Windows: python or C:\Python311\python.exe
   • macOS/Linux: python3
4. Configure timeout and confirmation preferences
5. Done — every python code block in chat will now have a Run Python button

Using your vault path in scripts

import os

vault = os.environ.get("VAULT_PATH", ".")
print(f"My vault is at: {vault}")

# List all markdown files in the vault
import pathlib
notes = list(pathlib.Path(vault).rglob("*.md"))
print(f"Found {len(notes)} notes")

---

🔧 Development

git clone https://github.com/disonfranco/compython
cd compython
npm install
npm run dev      # watch mode
npm run build    # production build

Copy main.js, manifest.json, and styles.css to your vault's plugin folder after building.

---

🗺️ Roadmap

• Auto-run Python after AI response (agentic loop for local models)
• Plot rendering — display matplotlib figures inline in chat
• Session state — persistent Python environment across executions
• Virtual environment support — point to a venv instead of system Python
• Safety sandbox — optional restricted execution mode

---

🙏 Credits

Compython is built on top of Smart Composer by @glowingjade and contributors. All original features and architecture are theirs — Compython adds the Python execution layer on top.

Authors of the Python execution feature:

• Dison Franco — concept, testing, and direction
• Claude (Anthropic) — implementation

---

📄 License

MIT — same as the original Smart Composer.