Back to directory
AI & Automationpungggi/pi-continual-harness
pi-continual-harness
Online self-improvement layer for pi. A unified, in-trajectory harness-state store (prompt notes, memory, skill descriptions, sub-agent specs) with a manual /refine that proposes evidence-backed structured CRUD deltas. Composes with pi-reflect (offline) a
Suggested install command
npx skills add pungggi/pi-continual-harness/pi-continual-harnessAlways 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 |