Back to directory
Docs & WritingIgorGanapolsky/rlhf-feedback-loop
rlhf-feedback-loop
Feedback-Driven Development (FDD) for AI agents — capture preference signals, steer behavior via Thompson Sampling, and export KTO/DPO training pairs for downstream fine-tuning.
Suggested install command
npx skills add IgorGanapolsky/rlhf-feedback-loop/rlhf-feedback-loopAlways inspect the linked repository and skill instructions before running commands. Skills are instructions; permissions and execution still matter.
Compatibility
Agent support matrix
3 supported
| Agent | Status |
|---|---|
| Claude Code | Supported |
| OpenCode | Not listed |
| Cursor | Supported |
| MCP | Not listed |
| GitHub Copilot | Not listed |
| Windsurf |