Create a Research¶
A research is DAAS's top-level artifact: a persisted bundle that ties an entity collection, an indicator collection, a dashboard, and a data pipeline to a generated markdown report. Use it for a company, an industry, a city, or a country.
The flow¶
goal -> brainstorm (optional) -> [entity collection] + [indicator collection]
-> indicators computed -> dashboard -> research bundle -> markdown report
1. (Optional) Brainstorm the goal¶
If your goal is fuzzy, start with fd-daas-brainstorm. It clarifies the goal
through dialogue and writes a plan to daas-doc/research/<plan-slug>.md - no
daas.db state. See fd-daas-brainstorm.
2. Run the research skill¶
The fd-daas-research skill orchestrates the whole flow. In Claude Code:
Run a research on BYD's trend: build the entity + indicator collections, compute the indicators, make a dashboard, and persist it as a research bundle with a markdown report.
The skill:
- Auto-detects prior researches and decides refresh-vs-rebuild.
- Builds (or reuses) the entity + indicator collections.
- Computes each indicator (
daas_run_indicator) and syncs rule-based collections. - Creates/updates a dashboard.
- Persists a
researchesrow + writesresearches/<name>.md. - Emits a
skill-run-notificationblock.
3. Run via MCP¶
If you prefer direct tool calls:
research_create(name="byd-trend",
entity_collection_name="byd-watchlist",
indicator_collection_name="byd-momentum",
dashboard_slug="byd-trend",
pipeline_collection_name="byd-pipeline")
research_generate_report(name="byd-trend")
Refresh the data later (recompute indicators, re-sync collections):
4. Inspect¶
Or via MCP: research_get(name="byd-trend") / research_list().
Works for any entity type¶
The same flow works for a company (BYD, AAPL), an industry (a collection of stocks), a city (a regional entity + its indicators), or a country (CN/US macro series). The collections you build define the scope.
Next¶
- See the dashboard you just made: Share a Dashboard.
- Understand the bundle: fd-daas-research.