What problem does it solve? Undocumented internal or third-party web APIs force developers to manually inspect network traffic and hand-write specifications. This Skill converts observed browser traffic (HAR files or live Chrome/Edge CDP sessions) into a complete OpenAPI 3.1 specification, an interactive HTML explorer, and a ready-to-use JavaScript client SDK, entirely offline with zero npm dependencies. ## Core Features & Use Cases - Traffic-to-Spec Pipeline: A five-stage pipeline (load, filter, normalize, infer, emit) pairs requests with responses, strips analytics noise, templatizes path parameters (UUIDs, numeric IDs, slugs), and infers JSON schemas from observed payloads. - Multiplexed Endpoint Decomposition: Automatically splits GraphQL and JSON-RPC endpoints into distinct named operations, including persisted-query hash extraction. - Multiple Input Methods: Convert an exported .har file directly, record a live local Chrome/Edge session via CDP, or consume HAR output from Playwright/Puppeteer scripts. - Privacy-First Output: Redacts sensitive headers, tokens, and PII before generating deliverables: openapi.yaml/json, index.html explorer, client.mjs SDK, report.md, and confidence.json. - Use Case: You need to integrate with a vendor's undocumented web app. Record a browsing session, run the discovery pipeline, and receive a full OpenAPI spec plus a callable client module. ## Quick Start Ask the agent to convert your exported session.har file into an OpenAPI specification and interactive API explorer using the browser-to-api skill.
