start_discovery
writeResearch runs
Start a research run (Discovery) for the project.
Connected surface
This tool and the Research dashboard page are two surfaces over the same data. The dashboard is where you view and curate it;
start_discovery is how your AI can launch a research session from inside the conversation that prompted it. How surfaces connect →Write tool
Mutates Rampify state, triggers background work, or calls a paid external API. Most clients ask for confirmation before invoking.
Parameters
| Parameter | Type | Description | |
|---|---|---|---|
| domain | string | optional | Project domain (alt to project_id). |
| project_id | string | optional | Project UUID. |
| product_id | string | optional | Optional. Product to scope this session to. Defaults to the project's default product when omitted. Use list_products to discover product IDs for multi-product projects. |
| playbook | enum("category_visibility" | "comparison_readiness" | "scenario_surfacing" | "developer_workflow" | "narrative_audit" | "market_opportunity") | optional | Optional. Map the user's natural-language playbook name to one of the slugs. Omit only when the user did not name a playbook at all — the tool will tell you whether to pick automatically or ask the user. |
| template | enum("category_visibility" | "comparison_readiness" | "scenario_surfacing" | "developer_workflow" | "narrative_audit" | "market_opportunity") | optional | Deprecated: the same as playbook. |
Related tools
derive_keyword_clusterswriteTurn research findings into pre-clustered keyword groups with live volume datalist_research_sessionsreadList research runs for the projectget_research_sessionreadGet session detail (summary by default, full responses with detail: "full")get_discovery_actionreadRead the current next_action for a research run (no side effects)