iterate — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited iterate (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Spin up N parallel agents on isolated git worktrees, each exploring a different direction for the same problem. Review all variants, pick the best, merge it back.
/iterate <task description>
/iterate 3 <task description> # explicit count (default: 3)
/iterate 2 redesign the settings page using shadcnParse $ARGUMENTS to extract:
If no arguments provided, ask the user:
Before spawning agents, generate N distinct approaches. These should be meaningfully different, not just cosmetic variations.
Example for "redesign the settings page":
Example for "implement caching":
Present the angles to the user for approval before spawning. The user may adjust, add, or remove angles.
For each variant, spawn an agent using:
Agent tool with:
- isolation: "worktree"
- subagent_type: "general-purpose"
- run_in_background: true (to run all N in parallel)Each agent gets this prompt structure:
You are variant {i} of {N} exploring: {task}
YOUR ANGLE: {angle_description}
INSTRUCTIONS:
1. Implement the solution following the angle described above
2. Make it complete and functional — this will be compared against other approaches
3. If this is a web project with a dev server, start it on port {base_port + i}
(use the project's dev command with --port flag)
4. When done, write a brief VARIANT_NOTES.md at the repo root with:
- Approach taken
- Key decisions and tradeoffs
- What you'd improve with more time
5. Commit all changes with message: "iterate: variant {i} — {angle_name}"Port assignment: Base port = 3100, so variants get 3101, 3102, 3103, etc.
Wait for all agents to complete. As each finishes, collect:
Present a comparison table:
┌─────────┬──────────────────────┬──────────┬─────────────────────────┐
│ Variant │ Angle │ Port │ Summary │
├─────────┼──────────────────────┼──────────┼─────────────────────────┤
│ 1 │ Minimal │ :3101 │ Single page, 3 sections │
│ 2 │ Tabbed │ :3102 │ 4 tabs, sidebar nav │
│ 3 │ Search-first │ :3103 │ Spotlight + categories │
└─────────┴──────────────────────┴──────────┴─────────────────────────┘Tell the user to review each variant (visit the ports, check the code).
Ask the user: "Which variant do you want to keep?"
Options:
If single winner:
git worktree remove)If cherry-pick:
If none:
Always clean up worktrees when done, even if the user abandons the process:
# List worktrees
git worktree list
# Remove each variant worktree
git worktree remove <path> --force
# Delete variant branches
git branch -D iterate-variant-1 iterate-variant-2 iterate-variant-3
# Kill dev servers on variant ports
lsof -ti:3101 -ti:3102 -ti:3103 | xargs kill 2>/dev/nullpackage.json scripts (dev, start, serve) or common framework commands to determine if/how to start a preview server~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.