Context Curator

Context Curator Skill

Purpose

Dynamically load relevant context based on the current task type. This skill implements the "Curation Layer" from context engineering principles.

When to Use

• Starting a new complex task
• Resuming work on a feature after a break
• Need to recall past decisions or patterns
• Want to ensure consistent patterns with previous work

Task Type Detection

Analyze the user's request to determine task type:

Execution Steps

Step 1: Detect Task Type

Analyze user request keywords to identify task type from the table above.

Step 2: Load Memory Context

Search Memory MCP for relevant entities:

mcp__memory__search_nodes({ query: "[task-type] [feature-name]" })

Key entities to query:

• ProjectContext - Overall codebase structure
• FeatureProgress_[branch] - Current feature status
• PatternHistory - Successfully used patterns
• RecentDecisions - Recent architectural decisions

Step 3: Suggest Relevant Skills

Based on detected task type, recommend specific skills from the table.

Step 4: Load Reference Files

For each task type, suggest key reference files:

Backend CQRS:

• src/PlatformExampleApp/PlatformExampleApp.TextSnippet.Application/UseCaseCommands/ - Command examples
• .github/skills/backend-cqrs-command/SKILL.md - Patterns guide

Frontend Angular:

• src/PlatformExampleAppWeb/apps/playground-text-snippet/ - Component examples
• .github/skills/frontend-angular-component/SKILL.md - Patterns guide

Debugging:

• .github/AI-DEBUGGING-PROTOCOL.md - Full debugging protocol
• .github/instructions/debugging.instructions.md - Quick reference

Step 5: Output Context Summary

Provide formatted summary:

## Context Loaded

**Task Type:** [detected type]
**Suggested Skills:** [skill-1], [skill-2]
**Memory Context:** [relevant entities found]
**Reference Files:** [key files to read]

**Previous Work:**

- [Relevant past decisions]
- [Patterns used successfully]
- [Current feature progress]

**Next Steps:**

1. [Recommended first action]
2. [Recommended second action]

Memory Integration

After completing significant work, store learnings:

mcp__memory__create_entities({
  entities: [{
    name: "Decision_[timestamp]",
    entityType: "ArchitecturalDecision",
    observations: ["Chose X because Y", "Pattern used: Z"]
  }]
})

Example Usage

User: "I need to add a new endpoint for task management"

Context Curator Output:

## Context Loaded

**Task Type:** backend-cqrs + api-integration
**Suggested Skills:** backend-cqrs-command, backend-cqrs-query, frontend-angular-api-service
**Memory Context:** Found FeatureProgress_task-management with 3 observations

**Previous Work:**

- Task entity already created (TaskItemEntity.cs)
- Using SaveTaskItemCommand pattern

**Next Steps:**

1. Check existing TaskItemController for patterns
2. Use backend-cqrs-command skill for new command
3. Update frontend API service