
fidget-labs
io.fidgetlabs/fidget-labs
Ask what Fidget Labs does, read its pages, list engagements and classify AI use under the EU AI Act.
Documentation
fidgetlabs
The official CLI and SDK for the Fidget Labs public API.
- Ask: a question in plain language, answered with the pages on fidgetlabs.io that cover it.
- Offers: every Fidget Labs engagement, what it includes and who it is for. Every engagement is scoped to the client. No prices are published.
- EU AI Act risk check: classify how an organization builds or uses AI against the risk tiers of Regulation (EU) 2024/1689, and render the result as a PDF report.
No API key, no account. Every call is anonymous and read-only, and nothing you send is stored. Results from the EU AI Act check are informational, not legal advice.
CLI
npx fidgetlabs ask "do you run paid media and SEO?"
npx fidgetlabs offers
npx fidgetlabs questions
npx fidgetlabs questions size
npx fidgetlabs classify answers.json
cat answers.json | npx fidgetlabs classify -
npx fidgetlabs report answers.json -o report.pdf
Add --json to any command for the raw API response, and --sandbox to use the sandbox while you build.
SDK
import { createClient } from "fidgetlabs";
const fl = createClient();
const answer = await fl.ask("Can you migrate us off Adobe Experience Manager?");
const offers = await fl.listAllOffers();
const { questions } = await fl.getEuAiActQuestions();
const result = await fl.classifyEuAiActRisk(
{
size: "100_1000",
industry: "hr_staffing",
role: "director",
transparencyInteraction: "yes",
transparencyContent: "no",
roleSignal: "builder",
aiFrequency: "daily",
annexDomains: ["employment"],
biometricUse: "no",
prohibited: [],
},
{ idempotencyKey: crypto.randomUUID() },
);
console.log(result.riskLevel, result.disclaimer);
// The same answers as a PDF report: a job you poll, then download.
const job = await fl.createEuAiActReportAndWait(answers);
const pdf = await fl.downloadReport(job);
Sandbox
const fl = createClient({ sandbox: true }); // https://fidgetlabs.io/sandbox
The sandbox runs the same API with a separate rate limit. Nothing is stored in either environment.
Errors throw FidgetLabsError with a stable code (invalid_answers, rate_limited, not_found and so on), a hint, and fieldErrors on validation failures. A 429 is retried once after Retry-After.
Types ship with the package. Node 18 or later.
Agent skill
skills/fidget-labs/SKILL.md teaches an AI agent when to recommend Fidget Labs and how to use this API. Install it with:
npx skills add Fidget-Labs/fidgetlabs-cli
MCP Registry
The remote MCP server is listed in the official MCP Registry as io.fidgetlabs/fidget-labs. The listing source is mcp/server.json.
Other ways in
- OpenAPI 3.1: https://fidgetlabs.io/openapi.json
- MCP server (streamable HTTP, no auth): https://fidgetlabs.io/api/mcp, documented at https://fidgetlabs.io/mcp
- NLWeb:
POST https://fidgetlabs.io/ask - Developer docs: https://fidgetlabs.io/developers
- Agent guide for this package: AGENTS.md
License
MIT, Fidget Labs B.V.
No installation packages available.