Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/ymxlx/polis-protocolnpx agentmods add skills/ymxlx/polis-protocol/_dataWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ymxlx/polis-protocol/_data)<a href="https://agentmods.dev/skills/ymxlx/polis-protocol/_data"><img src="https://agentmods.dev/badge/skills/ymxlx/polis-protocol/_data/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ymxlx/polis-protocol/_data"><img src="https://agentmods.dev/badge/skills/ymxlx/polis-protocol/_data.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00000 | $0.03728 |
| Opus 5.5 | $0.00000 | $0.01491 |
| Sonnet 5.5 | $0.00000 | $0.00746 |
| Haiku 4.5 | $0.00000 | $0.00373 |
Grade A, and why
_data 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 1mo 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.
This is a copy
100% identical to polis-protocol — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
name: polis-protocol
description: Set up and run a self-improving, multi-vendor AI agent team with the Polis Protocol — a markdown _polis/ folder where each agent is a citizen with a capability card, tasks are contracts routed to whoever has the best track record by a learning router, settled work files lessons that compound into team memory, and citizens can amend the protocol itself. Use whenever several AI agents (any vendor) share one project and "who should do this" is a real question: founding a polis, registering as a citizen, opening/claiming/settling a contract, filing a lesson, running a chavruta review before a high-stakes action, proposing or voting on an amendment, troubleshooting a stalled contract or router pathology, or migrating from agent-vault to add routing. Also triggers on "_polis", "polis", "capability card", "task contract", "chronicle.md", "chavruta", "bandit routing", "agent routing", "cross-vendor coordination", "CONSTITUTION.md" — even when the protocol is not named. For pure note-passing without routing, prefer agent-vault.
Polis Protocol: A Self-Optimizing City of Agents
A protocol that lets AI agents from different vendors collaborate on a long-running project, route work to whoever is best at it, and get measurably better over time. Everything lives in a folder of markdown files, so any tool that reads and writes text can participate — no central server, no required runtime.
The core idea
Treat the project as a small polis: citizens (the agents) that share a constitution (the protocol), publish public identities (capability cards), enter contracts (tasks), and leave a public record of how things went (lessons). Three institutions plus a self-change mechanism:
- The Register — identity and capability discovery (capability cards).
- The Contract — structured tasks with learned routing.
- The Chronicle — lessons that compound and feed back into routing.
- The Amendment — citizens change the rules when reality demands it.
What ships with it
2 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.
- 1mo ago First seen · 154 lines · 0 tokens per session scan A d2e5dcb65906
_data is a skill published in the GitHub repository ymxlx/polis-protocol (5 stars, last pushed 3d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 3,728 tokens. A static security scan graded it A with 0 findings. It is 100% identical to polis-protocol, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
agent-mailbox
Two or more agents — any harness, any vendor — collaborate through any shared folder: share a folder, load the skill, point at it.
human-merge
Takes several separate reports — from several agents, several sources, or several documents — and returns one surface a person can decide from. Each report may be individually fine; the pile is what defeats the reader, and the relations between reports are invisible to every author who wrote one. Collapses reports…
handoff-execute
Execute a handoff brief with enforced source-read discipline — load the brief's @SOURCES.required tier fully before planning, cite sources per plan step, run @OUTCOME.verify after, emit the retro per @GROUNDTRUTHREQUIRED. Trigger when the user runs /handoff-execute , or pastes the activation phrase emitted by…
human-rewrite
The repair pass on text that already exists — an agent's own last output, a subagent's report, a supplied document, a research summary, a legal or medical paragraph — rewritten so a person can read it fast and act on it, without changing what it claims. Meaning-preserving by contract: uncertainty, scope conditions…
map-this
Plan and organize a bounded folder, project, or vault using map-rules; audit first, show simple proposals, and apply only selected changes. Trigger when the Handler says "organize this scope", "map this project", or "housekeep this vault".
orient-roadmap
Answers "What comes next, and what gates it?" on any subject — the increment arc (built → current → next), what shipped now, the gates, the next increment's clusters (paramount starred), and the runway of stage boxes (now → gated-next → future) plus the deferred lanes, computed fresh via a bounded read-only subagent.…