<!-- ACE_SECTION v0.4.17 -->

ACE Pattern Learning Skill

This skill provides pattern-based learning for coding tasks.

When This Skill Activates

Copilot will automatically load this skill when your request contains:

• Implementation: implement, build, create, add, develop, write
• Modification: update, modify, change, edit, enhance, extend
• Debugging: debug, fix, troubleshoot, resolve, diagnose
• Refactoring: refactor, optimize, improve, restructure
• Integration: integrate, connect, setup, configure

BEFORE Starting Work (MANDATORY)

INVOKE:

#ce-dot-net.ace-vscode/ace_search

with relevant query

Search for patterns from previous work before implementing:

• Proven solutions and best practices
• Known gotchas and pitfalls to avoid
• Recommended APIs and libraries

AFTER Completing Work (MANDATORY)

INVOKE:

#ce-dot-net.ace-vscode/ace_learn

with:

• task

  : Brief description of what was accomplished

• success

  : true/false

• output

  : Key lessons, patterns discovered, gotchas

DO NOT skip this step. DO NOT respond to user without calling ace_learn first.

Mid-Conversation Re-Search

If user's follow-up message introduces a NEW topic or domain:

1. Recognize topic shift (auth→caching, API→database, etc.)
2. Call ace_search with the new topic BEFORE continuing
3. Apply newly retrieved patterns

Examples of topic shifts requiring re-search:

• "Now let's add caching" (after auth implementation)
• "I'm getting a database error" (after API work)
• "How do I deploy this?" (after coding)
• "Let's add tests" (after implementation)
• "Now handle the error cases" (new domain)

Example Workflow

User: "implement JWT authentication"
    ↓
1. ace_search("JWT authentication") → Find patterns
2. Implement using patterns found
3. ace_learn(task="Implemented JWT auth", success=true, output="Used refresh token rotation")
4. Respond to user

User: "Now add Redis caching for the tokens"
    ↓
1. ace_search("Redis caching tokens") → NEW SEARCH for new topic!
2. Implement caching using patterns found
3. ace_learn(task="Added Redis token caching", success=true, output="Used TTL matching token expiry")
4. Respond to user

<!-- ACE_SECTION_END -->