Copilot

Copilot Makes Obsidian the Knowledge Worker's IDE

Rewritten end to end. Copilot V4 puts frontier agents natively inside Obsidian.

No. 1 Obsidian AI Plugin

Install in Obsidian · Plans

Free with the AI you already pay for

Bring your own Claude or ChatGPT subscription and use Copilot for free. The plugin is open source, and each way of bringing your own AI maps to an agent that runs natively in your vault:

Prefer not to juggle API keys? The new Lite plan runs opencode on Copilot-hosted models with enterprise-grade privacy for $7.99 a month, and unlocks most paid features.

_{Provider terms and usage limits still apply.}

Rebuilt from the ground up

V4 is an entire rewrite. Instead of a single built-in assistant, real agents plan, search, use tools, and write results back to your notes.

• Speaks fluent Obsidian. Agents write wikilinks, canvases, and Markdown. The work lands in your vault and stays linked to everything around it.
• Tabs, like a browser. Open a different agent in each tab and work several threads at once: one summarizing a paper while another reorganizes your inbox.
• Project Mode, scoped to the work. Give an agent its own project, with a per-project AGENTS.md that steers how it behaves inside that project.
• Quick Ask, without leaving the page. Assign a shortcut in Hotkeys (we recommend Ctrl/Cmd+K) and trigger an inline, turn-by-turn window right from the note area, so you can ask a quick question without losing your place.
• Finds that vague thought instantly. Ask in half-remembered language and Copilot searches your vault by meaning, powered by Miyo, our local indexing engine.
• Any model, zero lock-in. OpenAI, Anthropic, Google, Ollama, LM Studio, or any OpenAI-compatible endpoint. Switch whenever you want.
• One setup, every agent. Add a skill once, then enable it across opencode, Claude Code, and Codex.

"Where's that note about attention only reading some of the tokens?"

Found it in [[Sparse attention]]: "Attend to a learned subset of tokens instead of the full window: quality holds while compute drops."

You also captured this in [[2026-08-03 Native sparse attention]] and [[2026-08-01 Longformer revisited]].

One prompt, every agent

[!IMPORTANT] Multi-agent collaboration is included with the Plus and Supporter plans.

Mention several agents in one prompt and each works it in parallel. Copilot combines their answers into a single result, so you get three perspectives for the effort of one question.

@claude @codex @opencode Review my draft on sparse attention and flag claims that newer papers contradict.

Pick your plan

Compare plans →

_{Copilot hosted models included in your license. Find your own usage on the token usage dashboard.}

Start in three steps

1. Install Copilot from Obsidian Community Plugins.
2. Open Copilot and choose opencode, Claude Code, or Codex.
3. Connect your subscription, API key, or local model and start working in your vault.

Trusted by people who think for a living

"The first tool that truly unifies how I search, organize, and retrieve knowledge without ever leaving Obsidian. My workflow is faster, deeper, and more connected. I can't imagine working without it."

Jason Zhang, Investor & Research Analyst

Frequently asked questions

Help and links

YouTube · Report a bug · Request a feature · Privacy

Support the project

If you share our vision for a powerful, portable AI agent for your second brain, consider sponsoring Copilot on GitHub or buying us a coffee.

Thank you to our GitHub Sponsors

Special thanks to our top sponsors: @mikelaaron, @pedramamini, @Arlorean, @dashinja, @azagore, @MTGMAD, @gpythomas, @emaynard, @scmarinelli, @borthwick, @adamhill, @gluecode, @rusi, @timgrote, @JiaruiYu-Consilium, @ddocta, @AMOz1, @chchwy, @pborenstein, @GitTom, @kazukgw, @mjluser1, @joesfer, @rwaal, @turnoutnow-harpreet, @dreznicek, @xrise-informatik, @jeremygentles, @ZhengRui, @bfoujols, @jsmith0475, @pagiaddlemon, @sebbyyyywebbyyy, @royschwartz2, @vikram11, @amiable-dev, @khalidhalim, @DrJsPBs, @chishaku, @Andrea18500, @shayonpal, @rhm2k, @snorcup, @JohnBub, @obstinatelark, @jonashaefele, @vishnu2kmohan

Paid Plan Disclosure

Copilot is a product of Brevilabs LLC and is not affiliated with Obsidian. Plus is one of its paid plan tiers. Visit obsidiancopilot.com for details.

• An account and payment are required for full access.
• Paid hosted models and cloud-backed features require network access.
• Privacy and data handling:
  • Free tier: Your messages and notes are sent only to your configured LLM provider, such as OpenAI, Anthropic, or Google. Nothing goes to Brevilabs servers.
  • Eligible paid tiers: Messages and attached note context go to the model provider you select. If you select an included Copilot-hosted chat model, that provider is Brevilabs and it receives the full prompt. If you select an included Copilot-hosted embedding model, Brevilabs receives the note text being indexed. Brevilabs also receives inputs for hosted features you invoke, including search queries, URLs, and files used by paid Quick Chat @ commands and Agent Mode project context.
  • Processing, not retention: We process data to deliver the feature you requested, then discard it. No message content, file uploads, or documents are retained on our servers after processing.
  • User ID: A randomly generated UUID is sent with hosted feature requests for service delivery, license abuse prevention, and rate limiting. It is not used for tracking, profiling, or analytics.
• See the privacy policy for more details.
• The Copilot plugin frontend is fully open source. Backend services that facilitate AI agents are closed source and proprietary.
• We offer a full refund within 14 days of purchase if you are not satisfied.

Authors

Brevilabs Team · [email protected] · @logancyang