Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/viktorbezdek/skillstack/swarm-protocolnpx skills add viktorbezdek/skillstack --skill swarm-protocolgit clone --depth 1 https://github.com/viktorbezdek/skillstackWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00218 | $0.03329 |
| Opus 5 | $0.00109 | $0.01665 |
| Sonnet 5 | $0.00044 | $0.00666 |
| Haiku 4.5 | $0.00022 | $0.00333 |
Grade A, and why
swarm-protocol scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 249 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Swarm Protocol
A persona swarm is parallel by design. Spawning 6 subagents one at a time defeats the point — they all see each other's work and converge. The value is in the dissent that emerges when each sees only the topic.
This skill owns the orchestration logic. It teaches when to invoke a swarm, how to choose the right persona subset, how to spawn them in true parallel via the Task() tool, and how to handle the returned outputs (with the synthesis handed off to swarm-synthesis).
⛔ MUST USE — Canonical persona names (literal subagent types)
These are the 12 literal subagent types shipped in agents/. They are NOT examples or suggestions — they are the exact strings you pass to Task(subagent_type=...). DO NOT invent your own persona names (no "Visionary", no "Pragmatist", no "Synthesizer", no "Bear Case Analyst", no "Risk Manager", no "Technical Expert" — those are not in the canonical set).
brainstorm-swarm:pm # Product Manager
brainstorm-swarm:engineer # Implementation Engineer
brainstorm-swarm:designer # UX / Product Designer
brainstorm-swarm:skeptic # Devil's Advocate
brainstorm-swarm:user-advocate # Customer voice
brainstorm-swarm:pre-mortem-specialist # Imagined-failure analyst
brainstorm-swarm:junior # Naive questioner
brainstorm-swarm:veteran # War stories / pattern-matching
brainstorm-swarm:first-principles-thinker # Strip back to fundamentals
brainstorm-swarm:constraint-setter # Scope discipline
brainstorm-swarm:optimist # Yes-and / 10x version
brainstorm-swarm:operator # Production reality
Always refer to personas by these literal names (pm, engineer, designer, skeptic, user-advocate, pre-mortem-specialist, junior, veteran, first-principles-thinker, constraint-setter, optimist, operator). When a topic needs a perspective the canonical 12 don't cover (CFO, Compliance Officer, etc.), use the custom-personas skill — but always preserve the canonical names for the canonical roles.
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 249 lines · 218 tokens per session scan A 14a3dccde7b9
swarm-protocol is a skill published in the GitHub repository viktorbezdek/skillstack (11 stars, last pushed 2mo ago), licensed MIT. It adds 218 tokens to every session and 3,329 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
gossipcat
Use when installing or setting up gossipcat multi-agent orchestration (parallel review, consensus, adaptive dispatch) in Claude Code or Cursor. Installs the gossipcat MCP server and hands off to gossipstatus() for all live rules.
Elephant Agent
Complete guide to what Elephant Agent is, how to use it from the CLI, how herd, wake, Personal Model, proactive curiosity, skills, and tools fit together, and how to explain Elephant Agent clearly to users, contributors, or evaluators.
awesome-skills-master
Master catalog orchestrator and autonomous installer for AI agents. Use when exploring, cloning, discovering, or installing skills, plugins, or MCP servers from awesome-skills without using interactive CLI tools. Guides environment detection, rule adaptation (.mdc/CLAUDE.md), MCP configuration, and strict author…
skills-sh-registry
Discover, query, evaluate, and fetch skills from the open skills.sh ecosystem (Vercel Labs) and global agent skill repositories.
agy-customizations
Comprehensive guide and reference for the Antigravity Customization System. Use to author skills, contextual rules, plugins, hooks, and MCP servers with correct priority loading.
context-mode
Context optimization and compression routing rules for AI agents exploring large codebases, reading massive logs, searching symbols, and batching tool executions.