← ClaudeAtlas

evidence-researchlisted

Investigate questions, compare options or assess gaps using source-grounded research and synthesis. Use when consequential claims require locating and reconciling evidence; skip simple fact lookups, text summaries and routine implementation.
Muratovnik/assay · ★ 0 · Code & Development · score 75
Install: claude install-skill Muratovnik/assay
# Evidence research Resolve the user's question with enough evidence to support a useful answer. This is a research method, not a new search service or an instruction to run agents. Use available domain tools and existing research artifacts. If another explicitly selected research workflow is active, contribute missing criteria within that workflow rather than starting a second plan or source ledger. ## Frame the work Identify the decision or explanation sought, material constraints, audience, time/version scope and desired deliverable. Reuse applicable prior evidence, checking volatile assumptions. Clarify only uncertainty that would materially change the work; distinguish a user's proposed solution from the underlying need. For a substantial investigation, keep a compact map of answer-bearing questions in the existing task or notes. Select families appropriate to the task, including consequential areas beyond reported examples. Update affected questions when the brief changes. Do not impose a full landscape survey on a narrow correction. ## Read the relevant procedure - For a landscape, a search gap, unfamiliar terminology or a claim of completeness, use [search and coverage](references/search-and-coverage.md). - For disputed claims, quantitative results, source chains or notes that will be reused, use [evidence and claims](references/evidence-and-claims.md). - For choosing options, explaining causes or generalizing a pattern, use [comparison and synthesis](refe