generated: '2026-07-21' method: searched status: published source: https://docs.traversal.com/using-traversal/mcp description: 'Traversal operates an official hosted (remote) MCP server that lets any streamable-http MCP client — Claude Code, Claude Desktop, Cursor, OpenCode, ChatGPT, Windsurf, VS Code — run investigations against a production stack. Over MCP, Traversal acts as a peer agent: the local agent handles code / terminal / kubectl while Traversal investigates across metrics, logs, traces, deployments and topology, handing executable next steps back. Agent-to-agent calls run in "Lightning mode" (a first pass in ~30s–2min). Tool list captured verbatim from the Traversal MCP documentation.' server: name: Traversal transport: http transport_detail: streamable-http url: https://api.prod.traversal.com/mcp/ auth: api_key auth_detail: 'Authorization: Bearer trv_ak_... — an API key generated in the Traversal web app under Settings > API keys (app.traversal.com/settings/api-keys).' self_hosted_note: Single-tenant SaaS and BYOC deployments use the organization's own API URL + /mcp/. clients: - Claude Code - Claude Desktop - Cursor - OpenCode - ChatGPT - Windsurf - VS Code tools: - name: investigate description: Run a fast first-pass (Lightning mode) investigation, typically 30s–2min. The agent passes the incident, what it can access locally, and how focused to be. - name: follow_up description: Continue an investigation using its session_id — drill into findings, verify a hypothesis, or feed back evidence the agent gathered locally. - name: get_investigation description: Get the status and report for an investigation, with a ui_url link to the web app. - name: list_investigations description: List recent investigations for your organization. - name: product_feedback description: Send feedback about the MCP experience to the Traversal team. deployment: mode: remote endpoint: https://api.prod.traversal.com/mcp/ verified: probed probe: wall checked: '2026-08-12' source: catalog MCP census