templates/skills/task-use/research/context-strategy/SKILL.md
Use for domain rules and business-flow research with source evidence; not for low-level code tracing
npx skillsauth add ozerohax/assistagents task-use-research-context-strategyInstall 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> <trigger>Need to understand business domain, terms, and rules</trigger> <trigger>Trace business logic across documents and code modules</trigger> <trigger>Align requirements with user personas, goals, and constraints</trigger> <trigger>Find inconsistencies between requirements and existing logic</trigger> </when_to_use>
<task_request> <principles> <principle>State the domain question and expected output (glossary, rules, flow)</principle> <principle>Include known sources (docs paths, code modules, or terms)</principle> <principle>Limit scope to one business area or feature per request</principle> <principle>Specify depth level when needed: standard, deep, expert</principle> <principle>Ask for evidence: cite document sections and code references</principle> </principles> <examples> <good> <task>Explain the billing domain: entities, rules, and key flows. Provide sources.</task> <why>Clear goal, scoped domain, evidence requested</why> </good> <good> <task>Find the rules that define account suspension and where they are enforced.</task> <why>Specific rule set and expected output</why> </good> <bad> <task>Understand the business</task> <why>Too broad, no concrete question</why> </bad> </examples> </task_request>
<research_principles> <principle>Start from domain documents and related code modules</principle> <principle>Build a glossary of terms and map them to entities</principle> <principle>Extract business rules with conditions, actions, and exceptions</principle> <principle>Trace rules to enforcement points in code and workflow</principle> <principle>Separate confirmed rules from assumptions</principle> <principle>Note conflicts or gaps between documents and implementation</principle> </research_principles>
<research_strategies> <strategy name="glossary-first" use_for="Unclear domain language"> <step order="1">Collect domain terms from docs, UI, and code</step> <step order="2">Map terms to entities, states, and metrics</step> <step order="3">Confirm definitions with sources</step> </strategy> <strategy name="rule-extraction" use_for="Business rules"> <step order="1">Identify rule statements and decision points</step> <step order="2">Normalize rules into if/then form with exceptions</step> <step order="3">Trace where rules are enforced in code and workflow</step> </strategy> <strategy name="flow-mapping" use_for="End-to-end processes"> <step order="1">List actors and triggers</step> <step order="2">Map the flow and key state transitions</step> <step order="3">Identify edge cases and failure paths</step> </strategy> </research_strategies>
<output_requirements> <requirement>Provide a concise summary of the domain findings</requirement> <requirement>List key rules and entities with sources</requirement> <requirement>Highlight contradictions, gaps, or open questions</requirement> </output_requirements>
<agent_limitations> <cannot>Edit or write files</cannot> <cannot>Make product decisions without user confirmation</cannot> <cannot>Remember previous sessions</cannot> </agent_limitations>
<depth_levels> <level name="standard">Glossary + key rules</level> <level name="deep">Rules, flows, exceptions, and enforcement points</level> <level name="expert">Conflicts, risks, and change impact analysis</level> </depth_levels>
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)