Get Perspective On

Extract the user's unique perspective on a topic from the knowledge base. Returns brief, focused insights (not full article). Can be called directly by user or via content agent in headless mode.

Usage

/get-perspective-on <topic or question>

Parameters

• topic or question: What perspective to retrieve
  • Specific topic
  • Question format
  • Comparison
  • Application

Examples

/get-perspective-on AI agent adoption barriers
/get-perspective-on Why do companies resist AI?
/get-perspective-on How dopamine relates to social media
/get-perspective-on What's contrarian about my AI views?

Workflow

1. Search Knowledge Base - read role: voice (contract: scope-mount): this is the user's perspective, so the read is pinned to core explicitly, never widened and never inherited. Books/ and Document Insights/ are what he has read, not his view - they do not belong in a perspective answer.

   BRAIN_READ_SCOPE=core resources/local-brain-search/run_search.sh "<topic>" --mode spreading --limit 8 --json
   BRAIN_READ_SCOPE=core resources/local-brain-search/run_connections.sh "<top note>" --json

   • Find 3-5 most relevant permanent notes
   • Look for contrarian or non-obvious angles
2. Synthesize Perspective
   • Brief format (1-3 paragraphs)
   • Focus on unique angles
   • Cite specific permanent notes
   • Include "why this matters"
3. Return Response
   • For direct user: Display perspective with citations
   • For a content agent headless: Return text for content use

Output Format

🧠 the user's Perspective: [Topic]

[1-3 paragraphs synthesizing unique insights from permanent notes]

Key insights:
- [Insight 1 from [[Note A]]]
- [Insight 2 from [[Note B]]]
- [Insight 3 connecting [[Note C]] and [[Note D]]]

Why this matters:
[Brief explanation of significance or application]

📝 Cited Notes:
- [[Note Title 1]]
- [[Note Title 2]]
- [[Note Title 3]]

Integration with a content agent

When a content agent calls via headless mode:

cd $PROJECT_ROOT
claude -p "/get-perspective-on 'AI adoption barriers'" --output-format json

Caller scope: this is the workstation invocation path - a content agent runs on the user's machine with an authenticated Claude Code CLI. A caller running inside a Trinity container cannot use claude -p (turn-scoped credentials, plus the ent#643 no-CLI-subprocess rule); it must call chat_with_agent("cornelius", "...") over Trinity MCP instead.

a content agent receives:

{
  "result": "[Perspective text with citations]",
  "total_cost_usd": 0.34
}

Quality Criteria

Good perspective: ✅ Cites 3-5 specific permanent notes ✅ Highlights contrarian or non-obvious angles ✅ Connects to broader themes ✅ Includes concrete examples ✅ Authentic your voice

Weak perspective (regenerate): ❌ Generic AI knowledge ❌ No specific note citations ❌ Surface-level analysis ❌ Lacks unique angles

Use Cases

• Quick content ideas for social posts
• Video script foundations
• Article brainstorming
• Perspective validation
• Connection discovery

Notes

• Keep brief (1-3 paragraphs)
• Focus on what's unique to the user's thinking
• Always cite permanent notes
• Emphasize non-obvious connections
• Cost: ~$0.30-0.40 per call

State Dependencies

Completion Checklist

• Knowledge base searched for topic
• 3-5 most relevant permanent notes identified
• Contrarian or non-obvious angles prioritized
• Perspective synthesized (1-3 paragraphs)
• Specific permanent notes cited
• "Why this matters" included
• Output formatted for user or a content agent consumption