naftiko: "1.0.0-alpha2" info: title: Toolhouse Agent Runs Bridge description: >- Pulls Toolhouse agent run history (run status, execution logs, and crucially MCP server logs) and bridges them into Naftiko's OpenTelemetry pipeline so a single observability dashboard shows Toolhouse agent activity alongside Naftiko capability calls. The MCP-server-logs surface is especially valuable — Toolhouse captures every MCP tool invocation per run. tags: - Naftiko - Toolhouse - Partnership - Observability - Agent-Runs - MCP-Logs - OpenTelemetry created: '2026-05-15' modified: '2026-05-15' binds: - namespace: toolhouse-env description: Toolhouse Platform API key. keys: TOOLHOUSE_API_KEY: TOOLHOUSE_API_KEY capability: consumes: - namespace: toolhouse type: http baseUri: https://api.toolhouse.ai authentication: type: bearer token: '{{TOOLHOUSE_API_KEY}}' resources: - name: list-agent-runs path: '/me/agent-runs' operations: - name: list-agent-runs method: GET inputParameters: - { name: next_cursor, in: query, type: string, required: false } - { name: limit, in: query, type: integer, required: false } - { name: agent_id, in: query, type: string, required: false } - { name: status, in: query, type: string, required: false } - name: agent-run path: '/me/agent-runs/{{run_id}}' operations: - name: get-agent-run method: GET inputParameters: - { name: run_id, in: path, required: true } - name: run-logs path: '/me/agent-runs/{{run_id}}/logs' operations: - name: list-run-logs method: GET inputParameters: - { name: run_id, in: path, required: true } - name: mcp-logs path: '/me/agent-runs/{{run_id}}/mcp-logs' operations: - name: list-mcp-logs method: GET inputParameters: - { name: run_id, in: path, required: true } exposes: - type: rest address: 0.0.0.0 port: 8080 namespace: toolhouse-agent-runs-bridge-rest description: REST surface for Toolhouse agent run observability. resources: - name: agent-runs path: '/agent-runs' operations: - name: list-agent-runs method: GET inputParameters: - { name: next_cursor, in: query, type: string, required: false } - { name: limit, in: query, type: integer, required: false } - { name: agent_id, in: query, type: string, required: false } - { name: status, in: query, type: string, required: false } call: toolhouse.list-agent-runs - name: agent-run path: '/agent-runs/{run_id}' operations: - name: get-agent-run method: GET inputParameters: - { name: run_id, in: path, type: string, required: true } call: toolhouse.get-agent-run - name: run-logs path: '/agent-runs/{run_id}/logs' operations: - name: list-run-logs method: GET inputParameters: - { name: run_id, in: path, type: string, required: true } call: toolhouse.list-run-logs - name: mcp-logs path: '/agent-runs/{run_id}/mcp-logs' operations: - name: list-mcp-logs method: GET inputParameters: - { name: run_id, in: path, type: string, required: true } call: toolhouse.list-mcp-logs - type: mcp address: 0.0.0.0 port: 3010 namespace: toolhouse-agent-runs-bridge-mcp description: MCP server for SRE / SOC / FinOps agents to query Toolhouse agent run history. tools: - name: list-agent-runs description: List Toolhouse agent runs with optional filters. hints: { readOnly: true } inputParameters: - { name: next_cursor, type: string, required: false, description: Pagination cursor. } - { name: limit, type: integer, required: false, description: Max runs per page. } - { name: agent_id, type: string, required: false, description: Filter to a specific agent. } - { name: status, type: string, required: false, description: Filter by run status. } call: toolhouse.list-agent-runs - name: get-agent-run description: Get a single Toolhouse agent run with status + metadata. hints: { readOnly: true } inputParameters: - { name: run_id, type: string, required: true, description: Toolhouse run UUID. } call: toolhouse.get-agent-run - name: list-run-logs description: List execution logs for a Toolhouse agent run. hints: { readOnly: true } inputParameters: - { name: run_id, type: string, required: true, description: Toolhouse run UUID. } call: toolhouse.list-run-logs - name: list-mcp-logs description: List MCP server logs for a Toolhouse agent run (which MCP tools were invoked, with what params, with what results). hints: { readOnly: true } inputParameters: - { name: run_id, type: string, required: true, description: Toolhouse run UUID. } call: toolhouse.list-mcp-logs