templates/skills/project/standart/research/SKILL.md
Run targeted evidence-based research in standard flow to close critical unknowns before architecture decisions
npx skillsauth add ozerohax/assistagents project-standart-researchInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
<when_to_use> <item importance="high">After brief, if critical unknowns or external constraints exist</item> <item importance="critical">Before architecture decisions in integrations, standards, or compliance contexts</item> </when_to_use>
<required_preload> <item>shared-base-rules</item> <item>shared-docs-paths</item> <item>web-strategy</item> <item>planning-risk-assessment</item> <item>planning-impact-analysis</item> </required_preload>
<document_target>
<rule importance="critical">Create a file in researches/{date time}-{user friendly name}.md</rule>
</document_target>
<output_format>
<section>Research questions</section> <section>Verified findings</section> <section>Assumptions / hypotheses</section> <section>Risks and constraints</section> <section>Impact on PRD/architecture</section> <section>Sources</section> </output_format><quality_rules> <rule importance="critical">Every key finding includes a source reference</rule> <rule importance="high">Unverified statements are marked as hypotheses</rule> <rule importance="high">Research outcomes change the next step, not remain "for reference"</rule> </quality_rules>
testing
Visual regression (baseline diffs, UI stabilization)
testing
Design detailed reproducible test cases (Given/When/Then, one check per case); not smoke/regression checklists
development
Mobile testing (native/hybrid/web, gestures, offline, interruptions)
testing
Load testing (k6/JMeter scenarios, benchmarks)