Radeon Arena Deploy Agent

Writes a plan for running an LLM locally on the user's AMD Radeon GPU, based on the user's request, the machine profile and setups measured on Radeon Arena, and answers errors hit while following it.

Skills

• Deployment plan
  Send {action:start, profile, need}; returns radeonrun.plan/v1 (ordered steps, endpoint) or radeonrun.need_info/v1 (questions). Answer questions or ask for changes with {action:answer, text}; a new plan comes back.
  amdradeonrocmllmdeploy
• Troubleshoot a step
  Send {action:report_issue, step_id, error_text}; returns radeonrun.fix/v1.
  troubleshooting
• Report outcome
  Send {action:report_result, success, note?} to close the session.
  feedback

How to call

https://radeon-arena.com/a2a

Listed in

Directories this entry was found in.