docs(runtime): reflect live reflex bodies + self-improvement loop #7

Merged
halofourteen merged 1 commits from docs/self-improvement-cycle into main 2026-05-25 16:22:39 +03:00
halofourteen commented 2026-05-25 16:22:35 +03:00 (Migrated from github.com)

Bring docs/runtime.md and README.md current with what's shipped through PR #5 and #6:

  • Reflex chain bodies (operator > defend > eat > sleep > idle) with the actual actions each dispatches.
  • Operator MC chat commands: status / come / pause / resume / stop.
  • Two Pi escalation paths: TUI hotkey `a` (manual) and the 20-noop auto-escalation, plus the rule that auto-escalation prompts must NOT propose code changes.
  • The end-to-end self-improvement loop with all three operator guardrails (approve → review → merge).
  • TUI hotkey `y` documented; `[proposals N, press y]` status-bar badge.
  • File layout updated with supervisor.js, actions.js, state-store.js, propose-apply.js.

🤖 Generated with Claude Code

Bring docs/runtime.md and README.md current with what's shipped through PR #5 and #6: - Reflex chain bodies (operator > defend > eat > sleep > idle) with the actual actions each dispatches. - Operator MC chat commands: status / come / pause / resume / stop. - Two Pi escalation paths: TUI hotkey \`a\` (manual) and the 20-noop auto-escalation, plus the rule that auto-escalation prompts must NOT propose code changes. - The end-to-end self-improvement loop with all three operator guardrails (approve → review → merge). - TUI hotkey \`y\` documented; \`[proposals N, press y]\` status-bar badge. - File layout updated with supervisor.js, actions.js, state-store.js, propose-apply.js. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
Sign in to join this conversation.