skills.rest
vercel-labs
vercel-labsOFFICIAL
Explore all skills by this author →
Website𝕏@vercelGitHub

derive-client

Generates standalone API clients from browser network traffic recorded in HAR files.

41.7k|2.8k|Updated Jan 11, 2026
Search Tags:#reverse engineering#api client#Browser Automation#agent-browser#har#network traffic#CLI generation
One-click install
npx skills add https://github.com/vercel-labs/agent-browser --skill derive-client
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: derive-client
Source: https://github.com/vercel-labs/agent-browser/tree/main/skill-data/derive-client
Command: npx skills add https://github.com/vercel-labs/agent-browser --skill derive-client

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Repeatedly driving a browser to automate the same website is slow and fragile. This Skill records a site's network traffic once, reverse-engineers its internal API from the captured requests, and generates a standalone client or CLI that calls the endpoints directly over HTTP.

Core Features & Use Cases

  • HAR-Based Recording: Captures browser traffic with embedded response bodies using agent-browser, preserving everything needed to study endpoints offline.
  • Endpoint Identification and Extraction: Uses jq queries to filter real JSON API calls from analytics noise, then extracts request shapes, response schemas, and auth material like cookies and CSRF tokens.
  • Client Generation and Verification: Produces one typed function per recorded flow with auth loaded from files or environment variables, then verifies every endpoint against the live API.
  • Use Case: You need to scrape a product catalog daily. Record one browsing session, derive the search and detail endpoints, and generate a CLI that fetches data directly without launching a browser again.

Quick Start

Record this website's network traffic with agent-browser and generate a standalone API client that calls its internal endpoints directly.

Frequently Asked Questions about derive-client

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I reverse engineer a website's internal API?▼

Record the site's network traffic with agent-browser network har start while using the site normally, then stop the capture to a HAR file. Query the HAR with jq to find JSON API endpoints, extract request shapes and auth headers, and write a client that replays those calls directly.

How to extract API endpoints from a HAR file?▼

Use jq to filter HAR entries by JSON mimeType and list method, status, and URL for each request. Ignore analytics endpoints like /collect or /track and third-party domains; the real API is usually first-party, JSON, and correlates with your recorded actions.

Why does my derived API client return 401 or 403 errors?▼

A 401 or 403 usually means the session expired or required headers are missing. Log in again through agent-browser, re-export cookies with agent-browser cookies get --json, and verify the client sends headers like user-agent, referer, or x-csrf-token that the API expects.

Can I automate a site without keeping the browser open?▼

Yes, after one recorded session the generated client talks to the site's internal API over plain HTTP with no browser needed. Only flows with per-session CSRF tokens or signed expiring parameters may still require the browser.

What are the limitations of deriving clients from recorded traffic?▼

Internal APIs are unversioned and can change without notice, so keep the HAR to re-derive the client. HAR files also contain live credentials like cookies and tokens, so treat them as secrets and keep them out of version control.

Related Skills

Explore More Software Engineering Skills→
fieldspherefieldsphere
37

browser-automation-tests

Automate browser UI flows and capture screenshots, console logs, and network data as evidence.

Official
Intermediate
spallempatispallempati

quartz-background-jobs

Generate scheduled background jobs with Quartz.NET for .NET 8+ applications.

Community
Advanced
cmetechcmetech

api-design

Design HTTP/GraphQL APIs with versioning, pagination, and error shapes.

Community
Advanced
AgentientAgentient
2

firebase-authentication-patterns

Implement Firebase Authentication flows with Google OAuth and cookie-based session management.

Official
Advanced