desk-research — guides¶

Evidence-grounded research, with the depth dialed to the question. /desk-research runs anything from a quick look-up to a deep investigation, grading its confidence as it goes. Around it sit the pipeline skills — build-outline, source-map, identify-perspectives, compare-hypotheses, devils-advocate, decision-archaeology — and two retrieval subagents that do the fetching.

The pack has two axes. Depth (above) is episodic — one-shot questions. Lifecycle is project mode: four research-project-* skills for a sustained, multi-week investigation that accumulates a corpus and ends in a brief you can hand to a decision. Episodic vs project research explains when to reach for which.

New here? Walk your first research session, then reach for the pipelines when one question needs several skills working together. For a sustained investigation, walk your first research project.

Tutorials¶

• Your first research session — install the pack and run /desk-research across all four depth modes.
• Your first research project — run the project lifecycle end to end: start, capture, digest, synthesize a brief.

How-to¶

• Run the research pipelines — the survey, decision, and archaeology recipes, and what each one produces.
• Run a research project and feed it into an RFC — drive the lifecycle, then promote the brief into an RFC's NNNN-notes/ companion and Evidence & prior art.

Reference¶

• Desk Research pack reference — every skill (episodic and project), the two retrieval subagents, the depth modes, the project folder layout and agentbundle-layout.toml's [research] table, and the GRADE confidence schema.

Explanation¶

• Research methodology — the why behind the pack — why depth is selectable, why sourcing comes before answering, and where confidence grades come from.
• Episodic vs project research — the two axes — depth vs lifecycle, the digest middle layer, and when a question deserves a project.

Installing, upgrading, and the adapter support matrix live in ../_shared/.