GitHub radar
Run several AI agents like a team — stop losing track of tasks
This is a ready-made set of instructions for your AI assistant: it keeps a task log from assignment to sign-off, so you stop losing track of who's doing what and what's actually done.
These are plain text instruction files ('skills') you copy into your AI assistant's skills folder so it keeps a log line for each task.
When it helps
Useful if you alone run several long-lived AI assistants at once. Skip it if you use one assistant for short, one-off tasks.
Pros
- One log line shows who took a task and who must accept it
- The method carries over between agents — Cursor, Codex, and others
- You can't mark a task done without real proof of the result
- Comes with a ready example and template line for your first try
Cons
- No built-in timers or reminders — you track deadlines yourself
- Docs are a draft, version 0.9, and may still change
- The only timing number covers one task in a pre-set Cursor folder, not a beginner's full setup
How to set it up — step by step
- 1Open the repo https://github.com/dabingfy-dot/solo-to-team-skills and download it
- 2Copy the skills/dispatch-accept-verify folder (with its templates) and the skills/evidence-before-done folder into your AI assistant's skills folder
- 3Open your AI assistant (Cursor, Codex, or another from the list) and reload its skills list
- 4Send the agent the text from the block below
- 5Walk through the example in examples/walkthrough-meeting-notes together with the agent
- 6Check the resulting log line: from 'waiting for someone' to 'accepted'
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up the Solo to Team skills from https://github.com/dabingfy-dot/solo-to-team-skills — copy the whole skills/dispatch-accept-verify folder (with its templates/) and skills/evidence-before-done into my AI agent's skills folder (Cursor, Codex, or another one listed in the README), run through the examples/walkthrough-meeting-notes example, and show me the resulting register line from pending to accepted.
The skills themselves are plain text files copied into the folder of an AI agent tool you already have installed (Cursor, Codex, and others from the list). No graphics card or extra memory needed — just a regular laptop where that agent already runs.
Open on GitHub▌ More finds