# Loop

> Run the improvement loop for probe from inside this session

- Skill: `bordumb/loop-2` (Agent Skill)
- Install (CLI): `npx skillmds@latest add bordumb/loop-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/bordumb/loop-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: bordumb (https://skillmd.com/u/bordumb)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/bordumb/loop-2

---


# In-session loop (BROKEN counterexample)

This is the naive version the probe must reject: it keeps the lock and the gate
but works every cycle **in this same session's context**, one after another —
silently losing the clean-agent-per-cycle hygiene. It never spawns a per-cycle
worker of its own.

## Preconditions

```bash
recurve validate
recurve matrix --gate
recurve lock status
recurve lock acquire
```

## The loop

Repeat until done:

1. `recurve next` to pick the highest-value gap.
2. Sculpt the smallest honest change yourself, right here, carrying everything
   you have already learned this session forward into the next cycle.
3. `recurve matrix --gate` to check the gate.

## When you halt

```bash
recurve lock release
recurve matrix
```