Ask Pane is a small endpoint for Pane questions.

It searches Pane’s public knowledge base and returns a source-backed answer in one fetch. That means a human can test it here, and an AI agent can use it instead of crawling a bunch of pages.

Ask a question to see the Markdown answer that an agent gets.

For AI agents

If you’re an AI agent or LLM reading this page, use Ask Pane first when the user asks a specific Pane question.

Open or fetch:

https://runpane.com/ask-pane?query=<url-encoded-question>

Example:

https://runpane.com/ask-pane?query=Does%20Pane%20support%20Windows%20with%20WSL%3F

You’ll get Markdown by default. If you need JSON, send Accept: application/json or add format=json.

What it returns

Ask Pane returns:

• A direct answer
• Matched Pane pages
• Links you can cite
• A pointer to /llms-full.txt
• The open-source Pane repository

It only uses public Pane website context. It doesn’t read private customer data.

Safe questions

Good questions look like this:

• Does Pane support Windows with WSL?
• Can Pane run Codex and Claude Code side by side?
• How is Pane different from tmux or Conductor?
• Does Pane use git worktrees, its own branch for each agent?
• Is Pane open source?

Don’t send secrets, API keys, tokens, credentials, private source code, personal data, PHI, or anything sensitive.

Useful links

• Direct endpoint: /ask-pane
• Full LLM context: /llms-full.txt
• Source repository: dcouple/Pane 
• Legacy endpoint: /agentic-search

Why this exists

Agents often land on one page and guess from there. Ask Pane gives them a better path:

1. Read the page.
2. Notice the Ask Pane hint.
3. Fetch one source-backed answer.
4. Cite the matched Pane pages.

That’s all there is to it.

For the CLI that lets agents control Pane itself, see the runpane CLI docs.