Skip to main content
Aquila ships a Model Context Protocol server. Connect any MCP client to Aquila, and the client can call the same recon tools that Hackbot uses. For step-by-step guides for Claude Code, OpenCode, Cursor, and Codex, read Connect a coding agent.

​
Endpoint and auth

The server speaks Streamable HTTP:
Send your MCP authorization key as a bearer token on every request:
The server validates the key on every request and scopes every tool call to your account. Your client sees only your targets, leads, and findings. A missing, wrong, or rotated key gets 401 Unauthorized.
The MCP key (aqmcp_...) is not your API key (aquila_...). The HTTP API accepts the API key in x-api-key. The MCP server accepts only the MCP key.

​
Get a key

Aquila creates your key when you sign up. Copy it from Settings → MCP Server in the dashboard: click Reveal, then Copy. The HTTP API also manages the key:
  • GET /api/mcp-key returns the key status.
  • GET /api/mcp-key/reveal returns the plaintext key when Aquila can still read it. Otherwise rotate first.
  • POST /api/mcp-key/rotate replaces the key. The old key stops working immediately.

​
Tool surface

The server groups its tools into four families. Your client lists every tool with its schema when it connects.
  • Recon. Enumeration: subfinder, passivesubs, crtname, chaos, relateddomains. DNS and TLS: dnsx, tlsx. External asset search: uncover, shodan, shodan_host, asn_lookup. Probing and crawling: httpx, katana, curl, ffuf. History: wayback. Wordlists: list_wordlists, wordlist_search.
  • Platform data. Reads of your own Aquila data: leads (search_leads, get_lead), vulnerabilities (search_vulnerabilities, get_vulnerability), scans and events (list_scans, get_scan, get_event_timeline), agent runs (list_agent_runs, get_agent_output), agent memory and inventory (search_memories, search_inventory), JavaScript evidence (search_javascript), alert channels (list_alert_channels), and your out-of-band endpoint (oob_endpoint, oob_interactions).
  • Platform actions. remember saves a learning to agent memory. save_javascript registers a captured script for monitoring. analyze_javascript queues an AI analysis pass over one. send_alert posts to one of your notification channels.
  • Bug bounty. Program discovery, scope resolution, and disclosed-report research on HackerOne and Intigriti: search_programs, fetch_programs, fetch_latest, get_program_scope, check_asset_in_scope, search_reports, and fetch_report. With your HackerOne credential the official Hacker Resources tools join these, from hackerone_account and structured_scopes to create_report and the Report Assistant draft tools.
Tools that touch a target enforce your approved scan scope before they send anything. An out-of-scope call is rejected and nothing leaves the server. The data and action tools touch only your own account data. The API reference documents the HTTP data endpoints the platform tools read from. Tool names and schemas follow the MCP standard and your client lists them at connect time.