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 skills add Goldziher/poly --skill poly-orchestratorgit clone --depth 1 https://github.com/Goldziher/polyWrote 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/goldziher/poly/poly-orchestrator)<a href="https://agentmods.dev/skills/goldziher/poly/poly-orchestrator"><img src="https://agentmods.dev/badge/skills/goldziher/poly/poly-orchestrator/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/goldziher/poly/poly-orchestrator"><img src="https://agentmods.dev/badge/skills/goldziher/poly/poly-orchestrator.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00043 | $0.01418 |
| Opus 5.5 | $0.00017 | $0.00567 |
| Sonnet 5.5 | $0.00009 | $0.00284 |
| Haiku 4.5 | $0.00004 | $0.00142 |
Grade A, and why
poly-orchestrator 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 — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
poly as the Orchestrator
Treat poly as the single lint and format gate for the repository. Do not invoke ruff,
oxlint, taplo, rumdl, or the rest directly — poly compiles them in as backends, and covers
everything else via the tree-sitter generic tier, the quality metric tier, and its built-in
ast-grep rule pack, behind one binary, one config, and one report.
Adopt it
- One
poly.toml. Configure every language and rule in a single per-repo file (poly.local.tomlfor local overrides,extendsto share a base). Layering is tool default → poly's opinionated overrides → yourpoly.toml. In a monorepo a nestedpoly.tomldeep-merges on top of its ancestors for the files beneath it. poly migrate— absorb existing tool configs intopoly.tomlinstead of hand-writing it. It imports ruff, typos, taplo, and markdownlint configs, and turns a lefthook or pre-commit config into[hooks],[hooks.builtin]and[tools]entries; there is no eslint or prettier importer (an existing.oxlintrc.json/.oxfmtrc.jsonis read directly, so it needs none). The default is a dry-run report — pass--writeto apply (--allow-dirtyon a dirty tree),--recursefor a monorepo,--verifyto re-run poly afterwards, and--strip-supersededto also removepyproject.tomlsections for tools ruff replaces. It deletes a source only when nothing in it was lost.--to-v2rewrites an existing v1poly.toml([lint.<lang>.<tool>],[rules],per-file-ignores, …) into the v2 schema ([engines.<id>],[workspace] run,per_file_ignores), keeping comments.poly hooks install— wire the git-hook shims sopolyruns the configured stages on every commit, replacing a.pre-commit-config.yamlor lefthook setup. It installs the hook types yourpoly.tomlconfigures;--hook-typepicks specific ones. A foreign hook already there is kept as<hook>.legacyand run before poly; a lefthook hook is kept but not run (--replace-lefthookdiscards it), and--overwritediscards any preserved hook.poly hooks uninstallrestores what was there before;poly hooks run [stage]runs a stage by hand (--all-filesfor the whole tree,-vto show passing jobs' output).- External tools — enable any catalog tool with
[tools.<name>] enabled = true(golangci-lint,actionlint,hadolint,helm,tflint,buf, …). Pinversionand setinstall = "auto"to have poly download a checksum-verified release when it is missing; by default it never downloads anything. poly config showprints the effective merged config;poly lock [--upgrade] [--check]pins symbolic remoteextendsrefs and Git hook sources intopoly.lock;poly config schemaprints the JSON Schema forpoly.toml. An unknown key in an engine table is anunknown-config-keywarning that suggests the nearest valid key (--verboselists every key the table reads).- Components. Only the Linux musl release binary is bundled (every engine linked in). Every
other release binary is a core build that downloads engines as signed components on first use
(
poly rulesis unavailable there):poly components list|status|sync|install, and--offline/POLY_OFFLINE=1forbids downloads.[components]inpoly.tomlsetsinstallandon_missing;[engines.typos] enabled = trueopts into spell-checking. poly doctorreports which poly is running, every poly on PATH, and the config in effect — the first thing to run when two machines disagree.
What ships with it
1 file 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 Changed · +1 lines 743ae74e3378
- 3d ago Changed · +6 lines 506cee40dc7c
- 9d ago Changed · +12 lines 7f4acda3fac2
- 23d ago Changed · +4 lines 7bd46640c602
- 26d ago Changed · +6 lines 845970a8cb5e
- 2mo ago Changed · +4 lines 350887e146e0
- 2mo ago First seen · 52 lines · 43 tokens per session scan A 2372831c4a6c
poly-orchestrator is a skill published in the GitHub repository Goldziher/poly (20 stars, last pushed today), licensed MIT. It adds 43 tokens to every session and 1,418 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-08-31.
poly-orchestrator compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| poly-orchestratorGoldziher/poly | Skill | 20 | today | MIT | A | 43 | Claude Code |
| ruff-docspledgeandgrow/pledge-skills | Skill | 10 | 3mo ago | MIT | C | 31 | Claude Code, Codex |
| ESLint for Test QualityPramodDutta/qaskills | Skill | 245 | 4d ago | MIT | A | 50 | Claude Code, Codex |
| uxskillmds/skillmd | Skill | 143 | 4d ago | MIT | A | 38 | Claude Code |
| research-paper-figure-skill-factoryskillmds/skillmd | Skill | 143 | 4d ago | MIT | A | 169 | Claude Code |
| refine-liveskillmds/skillmd | Skill | 143 | 4d ago | MIT | A | 103 | Claude Code |
| bpmn-generatorskillmds/skillmd | Skill | 143 | 4d ago | MIT | A | 202 | Claude Code |