# Zahif («زاحف») v1.0.0

> Web collector, interpreter and researcher for AI agents by Trusted Intelligence: REST API, remote MCP server,
> Claude Skill, SDKs and a Workers Service Binding. Arabic-first, works with or without LLMs, never fabricates, and
> returns the credits each call actually used. Compatible with the core Firecrawl v2 endpoints listed in the
> compatibility table (https://claude-crawler.trusted-intel.cc/docs/compatibility). Firecrawl is a trademark of its owner; Zahif is not affiliated with it.
> Private beta: no payments; trial keys of 500 credits come from the administrator.

Base URL: https://claude-crawler.trusted-intel.cc
Auth: Authorization: Bearer zh-<key>
Modes: fast (سريع), extended (مطوّل قليلاً), deep (مطوّل), custom (مخصص); default fast.

## Docs

- [Overview](https://claude-crawler.trusted-intel.cc/docs): what Zahif is and a quick start
- [API reference](https://claude-crawler.trusted-intel.cc/docs/api): every endpoint with examples
- [OpenAPI 3.1](https://claude-crawler.trusted-intel.cc/openapi.json): machine-readable contract
- [Modes and settings](https://claude-crawler.trusted-intel.cc/docs/modes): work modes and every switch
- [MCP and agents](https://claude-crawler.trusted-intel.cc/docs/mcp): claude.ai, Claude Code, Claude Desktop, Cursor, VS Code, Kimi
- [Compatibility table](https://claude-crawler.trusted-intel.cc/docs/compatibility): which Firecrawl v2 endpoints and options are supported, partial or unsupported
- [Agent browser](https://claude-crawler.trusted-intel.cc/docs/browser): sessions with placement / display / presence / control switches, aria snapshots, live view
- [Device pairing](https://claude-crawler.trusted-intel.cc/docs/devices): use the user's real browser through the extension or zahif-relay
- [Service Binding](https://claude-crawler.trusted-intel.cc/docs/service-binding): calling Zahif from other Cloudflare Workers
- [Webhooks](https://claude-crawler.trusted-intel.cc/docs/webhooks): events and HMAC-SHA256 signatures
- [Errors and limits](https://claude-crawler.trusted-intel.cc/docs/errors): error codes, budgets, politeness

## Endpoints

- [POST /v2/scrape](https://claude-crawler.trusted-intel.cc/docs/api#scrape): one URL to markdown, html, links, images, screenshot, summary, json, branding, changeTracking
- [POST /v2/search](https://claude-crawler.trusted-intel.cc/docs/api#search): multi-source web/news/images search with optional scraping of hits
- [POST /v2/map](https://claude-crawler.trusted-intel.cc/docs/api#map): discover a site's URLs, ranked by a search term
- [POST /v2/crawl](https://claude-crawler.trusted-intel.cc/docs/api#crawl): background crawl; poll GET /v2/crawl/{id}
- [POST /v2/batch/scrape](https://claude-crawler.trusted-intel.cc/docs/api#batchScrape): many URLs in one job
- [POST /v2/extract](https://claude-crawler.trusted-intel.cc/docs/api#extract): structured JSON from URLs or the web by schema/prompt
- [POST /v2/agent](https://claude-crawler.trusted-intel.cc/docs/api#agent): research agent with numbered citations; poll GET /v2/agent/{id}
- [GET /v2/team/credit-usage](https://claude-crawler.trusted-intel.cc/docs/api#creditUsage): remaining credits and plan usage
- [POST /v2/zahif/browser/sessions](https://claude-crawler.trusted-intel.cc/docs/api#browserCreate): agent-browser session; then POST /v2/zahif/browser/sessions/{id}/actions
- [GET /health](https://claude-crawler.trusted-intel.cc/health): service health

## MCP

- [MCP connection guide](https://claude-crawler.trusted-intel.cc/docs/mcp): endpoint https://claude-crawler.trusted-intel.cc/mcp (POST, Streamable HTTP, Authorization: Bearer header)

## Optional

- [Full text for agents](https://claude-crawler.trusted-intel.cc/llms-full.txt): all docs in one file
- [Web platform](https://claude-crawler.trusted-intel.cc/app): playground, jobs, usage, settings
- [Bot information](https://claude-crawler.trusted-intel.cc/bot): TrustedIntelBot user agent and robots policy