Code Intelligence MCP Server

# Document Research Prompt ## Task **User Request:** {{task_description}} ## Target Documents {{docs_path}} ## Instructions 1. Use Glob to list and understand the structure under docs_path 2. Respect the include/exclude settings from `document_search` in context.yml 3. Read only highly relevant files, focusing on necessary sections 4. Extract rules to follow for this modification/investigation and sections the main agent should read 5. Focus on reference locations (file:line / file:line-range) in the output ## Output Format ### rule_refs Reference locations for rules to follow: - [Rule summary]: path:line ### must_read Sections the main agent must read: - [Purpose/Reason]: path:line-range ### dependencies Dependencies/impacts to consider: - [Target]: reason ### warnings Caveats/pitfalls: - [Content]: basis or reason ## Guidelines - Read the minimum necessary (no full file reads) - Always specify references in file:line format - Prioritize task-specific rules over general guidelines - If none found, explicitly state "No applicable rules" or "No required reading" - Priority order: 1. Architecture constraints 2. Data/schema requirements 3. Naming/placement conventions 4. Integration patterns 5. Security requirements