chieflab_request_specialist
ActiveTool of ChiefLab
P87 — delegate to a specialist agent. USE WHEN the launch needs more than copy: a graphic brief (kind='design'), a demo video shot list (kind='video'), a prospective-customer research plan (kind='research'), personalized outreach drafts (kind='outreach'), landing-page SEO recommendations (kind='seo'), or a verdict + recommended next move from measurement data (kind='analytics'). Specialists are deterministic stubs today — they return structured briefs / plans / drafts the founder can act on immediately. Future LLM-backed implementations swap in transparently. Each request creates an approval-gated action so the brief lands in the Launch Room. Pass runId for grounding in the originating launch's intelligence brief.
Parameters schema
{
"type": "object",
"required": [
"kind"
],
"properties": {
"kind": {
"type": "string",
"description": "design | video | research | outreach | seo | analytics"
},
"brief": {
"type": "object",
"description": "kind-specific brief fields (asset/channel for design; durationSeconds/platform for video; goal for research; contacts/channel for outreach; url/targetTerms for seo; metrics/channel for analytics)"
},
"runId": {
"type": "string",
"description": "Optional. Source run id; the specialist grounds in that launch's brief (productName / audience / market / positioning)."
},
"workspaceId": {
"type": "string",
"description": "Optional workspace id."
},
"autoCreateAction": {
"type": "boolean",
"description": "Default true. When true, creates an approval-gated action in actionStore so the brief lands in the Launch Room. When false, returns the specialist output without persistence (use for one-off / preview)."
}
}
}No endpoints wrapped at confidence ≥ 0.50.
Parent server
ChiefLab
https://github.com/bdentech/chiefmo-agent
1/7 registries