loop

/loop — Bounded Repetitive Task Execution

Run a task across multiple targets (files, modules, endpoints, etc.) with safety bounds, progress tracking, and stall detection.

Flags

Flag	Default	Description
--targets	(required)	Comma-separated list or glob pattern of targets
--parallel	false	Run iterations concurrently (use with caution)
--max	10	Maximum number of iterations (safety bound)
--continue-on-error	false	Continue to next target if one fails

Instructions

Step 1: Parse Arguments

From $ARGUMENTS, extract:

• Task description: What to do to each target
• Targets: List of files, modules, or items to iterate over
• Execution mode: Sequential (default) or parallel
• Max iterations: Safety cap (default: 10)
• Error handling: Stop on first error (default) or continue

Step 2: Resolve Targets

If targets are a glob pattern, resolve them:

# Resolve glob to file list
ls [glob-pattern] 2>/dev/null

If targets are a comma-separated list, split them.

Validate:

• Target count must be > 0 (error if empty)
• Target count must be <= max iterations (warn and cap if exceeded)
• Each target must exist (if file-based)

Step 3: Initialize State

Create .planning/loop-state.json:

{
  "task": "[description]",
  "mode": "sequential|parallel",
  "max_iterations": 10,
  "targets": ["target1", "target2"],
  "status": "running",
  "started": "[ISO timestamp]",
  "iterations": [],
  "stall_count": 0
}

Step 4: Pre-Loop Baseline

# Capture baseline test status
npm test 2>&1 | tail -5
git status --short

Record baseline state for stall detection.

Step 5: Execute Loop

For each target (sequential or parallel based on mode):

1. Announce: "Iteration [N]/[total]: Processing [target]"

2. Execute the task on the current target

3. Verify iteration:

# Run tests after each iteration
npm test 2>&1 | tail -5