Name: He Research
Author: mattjefferson

Resolve investigatable unknowns before implementation planning.

Use this skill when answers are discoverable through research. For unknowns that must be built and experienced, use he-spike instead.

When to Use

Key Principles

Categorize first — only research questions where the answer can be found.
Evidence-backed — record confidence and source notes; separate fact from inference.
Update the source of truth — write findings into docs/specs/<slug>-spec.md with revision notes.
Prefer primary sources — repo evidence and official docs beat summaries.
Do not plan here — research clarifies constraints; planning is .

Resolve investigatable unknowns before implementation planning.

Use this skill when answers are discoverable through research. For unknowns that must be built and experienced, use he-spike instead.

Categorize first — only research questions where the answer can be found.
Evidence-backed — record confidence and source notes; separate fact from inference.
Update the source of truth — write findings into docs/specs/<slug>-spec.md with revision notes.
Prefer primary sources — repo evidence and official docs beat summaries.
Do not plan here — research clarifies constraints; planning is .

Anti-Pattern	Better Approach
Guessing when evidence is thin	Mark as unresolved with confidence level
Researching questions that need hands-on exploration	Redirect to `he-spike` for experience-dependent unknowns
Adding implementation details to the spec	Keep implementation in `he-plan`; spec is intent only
Treating inference as fact	Always distinguish and label confidence explicitly
Researching in the main thread instead of parallel	Launch one subagent per question for concurrent investigation