tune-multi-agent-swarm — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited tune-multi-agent-swarm (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.
| Size | Agents | Quality | Cost | Latency | Best For |
|---|---|---|---|---|---|
| Minimal | 3 | Good | Low | Fast | Simple tasks |
| Standard | 5-7 | High | Medium | Medium | General-purpose |
| Large | 10+ | Highest | High | Slower | Complex, reliability-critical |
Rule: Start with 3 agents. Add more only if consensus quality < 80%. Each additional agent costs ~$0.02-$0.05 per task in token usage.
| Parameter | Default | Range | Impact |
|---|---|---|---|
| Bid timeout | 5s | 2-10s | Shorter = faster, may miss good bids |
| Min bids required | 2 | 1-5 | More = better selection, slower |
| Scoring formula | confidence × accuracy | Custom | Weight past performance |
| Winner cooldown | None | 0-3 tasks | Prevents monopolization |
| Bid diversity bonus | None | 0-20% | Encourage different approaches |
| Method | Quorum | Quality | Speed | Cost |
|---|---|---|---|---|
| First-past-threshold | 1 vote > 0.9 | Medium | Fastest | Lowest |
| Majority vote | 50%+1 | High | Medium | Medium |
| Supermajority | 67%+1 | Very High | Slower | Higher |
| Weighted expert vote | N/A | Highest | Medium | Medium |
Tuning: Start majority. If quality insufficient → weighted expert. If cost too high → first-past-threshold.
| Role | Default | Cost-Optimized | Quality-Optimized |
|---|---|---|---|
| Researcher | gpt-4o-mini | gpt-4o-mini | gpt-4o |
| Analyst | gpt-4o | gpt-4o-mini | gpt-4o |
| Writer | gpt-4o-mini | gpt-4o-mini | gpt-4o |
| Validator | gpt-4o-mini | gpt-4o-mini | gpt-4o-mini |
Cost impact: All gpt-4o (~$0.15/task) vs mixed (~$0.05/task) = 3x savings.
| Component | 3-Agent Swarm | 7-Agent Swarm |
|---|---|---|
| Bidding (all agents evaluate) | $0.01 | $0.025 |
| Execution (winners only) | $0.03 | $0.05 |
| Voting (all agents) | $0.01 | $0.025 |
| State coordination | $0.001 | $0.001 |
| Total | ~$0.05 | ~$0.10 |
Monthly estimate (5K tasks/day):
After tuning, compare:
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.