OpenYOLO

by kamijoucen
5
4
3
2
1
Score: 61/100

Description

obsidian ai 助手

Reviews

No reviews yet.

Stats

3
stars
1,407
downloads
0
forks
110
days
2
days
2
days
0
total PRs
0
open PRs
0
closed PRs
0
merged PRs
0
total issues
0
open issues
0
closed issues
1,633
commits

README file from

Github

OpenYOLO

English | 中文

Chat with AI in the Obsidian sidebar to explore, organize, and edit your notes.

OpenYOLO works on Obsidian desktop and requires opencode to be installed and configured.

What you can do

  • Ask about your notes: automatically include the current note, or attach multiple notes, external text files, and images. Pasting images is supported.
  • Read and edit notes: ask AI to summarize content, organize information, or update notes. See its actions and file changes, and allow or reject permission requests when they appear.
  • Follow task progress. Read replies as they arrive, expand or collapse reasoning, and track a todo list.
  • Browse subagents. With OpenCode 2, choose Open subagent session on the task’s tool card to view live output, tool calls, and history in a read-only view. The task status updates on the same card.
  • Keep chatting during background tasks. Once the current turn in the main conversation ends, you can send another message while its subagents are still running. If the main agent sends a follow-up on its own, it appears with the Background follow-up label.
  • Choose your model and mode: search your configured models, adjust supported reasoning levels, and switch between plan and build modes.
  • Pick up where you left off: reopen your latest conversation automatically or switch to an earlier one from the history list.
  • Fork a conversation. Use the fork button in the header to copy the entire current conversation history into a new session. This requires ACP fork support from the backend.
  • Reuse previous prompts: press ↑ at the start of the input or ↓ at the end to recall text. The latest 32 inputs are shared across sessions and saved locally across restarts. Editing the text resets browsing to the newest entry.
  • Save conversations as notes: export a conversation to Markdown in a folder of your choice, without overwriting existing notes.
  • Make it your own: type / to select commands or skills. Optionally enable and edit the note-assistant instructions in settings; enabling them writes a section to your vault's AGENTS.md.

How it works

OpenYOLO provides the chat interface inside Obsidian and does not include an agent implementation. It connects to a local opencode backend through the Agent Client Protocol (ACP). opencode handles model calls, tool execution, and session management.

Obsidian / OpenYOLO ←→ ACP ←→ opencode ←→ AI model

Get started

For manual installation, download main.js, manifest.json, and styles.css from the latest release. Place them in .obsidian/plugins/openyolo/ in your vault, then enable OpenYOLO under Settings → Community plugins.

Install opencode and run opencode auth login to configure a model provider. You're ready to start chatting.

If OpenYOLO cannot find opencode, set its executable path in the plugin settings.

Access and permissions

OpenYOLO starts the local opencode executable with the vault as its working directory. The process runs with your OS account permissions and applies your opencode tool and model-provider settings. It receives your prompts and attached content.

OpenYOLO uses Node.js to locate the executable and check that ACP file requests stay inside the vault, including when paths pass through symbolic links. The note picker searches Markdown paths locally and attaches the files you choose. The chat reads images when you paste them into the composer and writes Markdown to the clipboard when you click a diagram or formula copy button.

About

Originally forked from obsidian-yolo. OpenYOLO is now developed independently and does not maintain upstream compatibility.

MIT License