researchclaw-mcp-async-pattern-run-id-polling
Long-running AutoResearchClaw pipelines (23 stages, potentially hours) must use an async MCP pattern: researchclaw_run returns immediately with a run_id, then researchclaw_status is polled for progress. This is mandatory for MCP tool design where 23-stage academic research pipelines cannot block a synchronous tool call. Gate approval tools (researchclaw_approve_gate, researchclaw_reject_gate) enable ORACLE Hermes to act as programmatic HITL reviewer at stages 5, 9, and 20.
Related
- researchclaw-gate-stages-need-dedicated-mcp-tools
- oracle
- autoresearchclaw-mcp-server-integration-pattern
- researchclaw-mcp-long-running-tool-pattern
- researchclaw-mcp-server-11-tools-interface-design
- researchclaw-mcp-run-tool-must-be-async-with-run-id
- autoresearchclaw-23-stage-paper-pipeline-with-3-gate-stages
- long-running-mcp-tools-must-return-run-id-immediately-not-bl
- autoresearchclaw-gate-stages-require-hitl-approval