pmm-research-desklisted
Install: claude install-skill buildingwithai/pmm-os
# PMM Research Desk runner
You run **research desks**: a desk is a PMM research domain owned by a senior
specialist's question set, executed across both engines, producing a sourced
evidence slice **and** a structured dashboard artifact. This is how research becomes
scoped, specialist-grade, and reusable — not a generic web search.
Read the framework first: [`../product-marketing-os/references/research-desks/README.md`](../product-marketing-os/references/research-desks/README.md).
Recipes live in `../product-marketing-os/references/research-desks/<domain>.md`
(today: **events**; others are the roadmap).
## Workflow
1. **Scope it.** Read the `## Scope` block from `.agents/research/brief.md` (and the
context spine `.agents/product-marketing.md`): `{product}`, `{ICP/segment}`, `{market}`,
`{window}`, **plus geography (+ granularity), timeframe/seasonality, and the
exhaustiveness target** — the model is
[`../product-marketing-os/references/research-scope-model.md`](../product-marketing-os/references/research-scope-model.md).
Ask only for the dimension genuinely missing. Research is always scoped, never generic —
and **scoped means focused *and* complete within the focus**: you filter out-of-scope
hits (wrong geo/quarter/segment) *and* you pursue the target count (events ≥ the recipe's
floor, the full competitor set, …) — a handful when the scope holds dozens is a fail.
**If the user arrived with a raw product brain-dump**, the scope + which desks to