run-autolisted

# /rein:run-auto Repeats the **per-task procedure of `/rein:run`** in a **bounded** loop, in one session, until an explicit stop condition. It does not reimplement selection, verification or closure — every iteration is a full `/rein:run` pass. Usage: `/rein:run-auto [change-name] [max-tasks]` `max-tasks` is the per-pass cap (default **3**). It bounds this invocation only. It is never unlimited. ```bash R=$(command -v rein || ls -d ~/.claude/plugins/cache/*/rein/*/bin/rein 2>/dev/null | sort -V | tail -1) ``` ## Loop 1. **Record this invocation** — never blocks, never fails the run. Shell state does NOT persist between tool calls, so `R` is resolved and used in the SAME block or it is empty and nothing is recorded: ```bash R=$(command -v rein || ls -d ~/.claude/plugins/cache/*/rein/*/bin/rein 2>/dev/null | sort -V | tail -1); "$R" event run-auto ``` 2. Set `done = 0`, `cap = max-tasks or 3`. 3. **Check the deterministic gate before each iteration**: `"$R" next .` 4. **Stop now, without claiming anything**, if any stop condition below holds. 5. Otherwise run **one full iteration of `/rein:run`**, unabridged: select, load only what the task needs, plan, implement in scope, run its verification plus the configured checks, compare against acceptance, close only with evidence. 6. Increment `done`. Re-check the conditions. Otherwise go to 3. ## Stop conditions — deterministic, not model judgement Each is read from a verifiable signal, **never** from you