War room research with MCP
Baseline first · candidate experiment
Enterprise tools provide internal facts first. Research only a genuinely external evidence question and join it locally; public research does not reconstruct internal company data.
Enterprise HTTP MCP + Parallel Responses · Parallel MCP unavailable on this connection
Main dependency
Actual vendor names, adapter contracts and coverage are unverified here; missing setup is not a claimed business gap.
- 01
Enterprise tools
Existing baseline · simulated here
- 02
Coverage check
Answered already? Stop; zero Parallel calls
- 03
Precise extra question
Optional, explicit action only
- 04
Incremental evidence
Supported additions, coverage and reviewer effort
Scenario input Synthetic
Outbound public context preview
{
"question": "What public manufacturer notices or repair guidance might explain delays for this vehicle family, and what needs direct shop confirmation?",
"public_vehicle": {
"year": 2020,
"make": "Toyota",
"model": "Camry"
},
"region": "Richmond Virginia"
}Kept local: internal_context (delayed_cases, reported_delay_days, private_canary)
Incremental outside research Optional
Enterprise HTTP MCP + Parallel Responses · Parallel MCP unavailable on this connection
A genuinely external evidence question, joined to internal context locally. Candidate experiment, not a verified deficiency. If existing evidence answers the question, stop.
No runs yet.