“BeeRad”andClaude Opus 4.5
2cbc950fd4
From private repo: Idea Genealogy: - Required edge explanations across UI/tools/API/MCP - Inferred edge schema (category/type/confidence) - Added part_of relationship type - Connection UI chips (Made by / Part of / Came from / Related) Quick Capture Overhaul: - Jina.ai fallback for JS-rendered sites (Twitter, SPAs) - Simplified UI: removed mode buttons, auto-detect input type - Renamed to "Add Stuff", moved to top-right - Auto-edge creation from description entities - Increased content summary length (500→1500 chars) Description Generation Fix: - Simplified prompt to "what is this" - Added link/dimensions context - Better creator attribution Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
68 lines
2.2 KiB
TypeScript
68 lines
2.2 KiB
TypeScript
export const INTEGRATE_WORKFLOW_INSTRUCTIONS = `You are executing the INTEGRATE workflow for the currently focused node.
|
|
|
|
MISSION
|
|
Find meaningful connections across the user's knowledge graph, create edges, and append an Integration Analysis.
|
|
|
|
YOU HAVE DIRECT WRITE ACCESS via updateNode (appends only) and createEdge (creates graph connections).
|
|
|
|
WORKFLOW STEPS
|
|
|
|
1. RETRIEVE & UNDERSTAND
|
|
- Call getNodesById for the focused node
|
|
- Identify: what type of thing is this? (person, project, paper, idea, video, etc.)
|
|
- Extract key entities: names, projects, concepts, techniques
|
|
- Note the core insight in one sentence
|
|
|
|
2. SEARCH FOR CONNECTIONS
|
|
Search the database using entities from step 1:
|
|
|
|
a) Structural connections:
|
|
- Names mentioned → queryNodes to find nodes about those people
|
|
- Projects/tools mentioned → queryNodes to find those nodes
|
|
|
|
b) Thematic connections:
|
|
- Use searchContentEmbeddings with key concepts
|
|
- Look for shared themes, complementary ideas, contradictions
|
|
|
|
Target: 3-5 strong connections (quality over quantity)
|
|
|
|
3. CREATE EDGES
|
|
For each connection found, call createEdge:
|
|
- from_node_id: the focused node ID
|
|
- to_node_id: the connected node ID
|
|
- explanation: "why this connection matters"
|
|
Direction rule: write the explanation so it reads FROM → TO.
|
|
Examples:
|
|
- Insight → Source: "Came from / inspired by"
|
|
- Episode → Podcast: "Episode of this podcast"
|
|
|
|
The tool handles duplicates gracefully - if edge exists, it returns an error and you continue.
|
|
|
|
Create 3-5 edges total.
|
|
|
|
4. DOCUMENT IN CONTENT
|
|
Call updateNode ONCE with ONLY this new section:
|
|
|
|
---
|
|
## Integration Analysis
|
|
|
|
[2-3 sentences: what this is, why it matters, core insight]
|
|
|
|
**Connections:**
|
|
- [NODE:123:"Title"] — [why connected]
|
|
- [NODE:456:"Title"] — [why connected]
|
|
[list all edges created in step 3]
|
|
|
|
CRITICAL: Send ONLY the new section. The tool appends automatically.
|
|
|
|
5. RETURN SUMMARY
|
|
Reply with: Task / Actions / Result / Nodes / Follow-up (≤100 words)
|
|
|
|
RULES
|
|
- Keep total tool calls ≤ 12
|
|
- Create edges BEFORE documenting (step 3 before step 4)
|
|
- Call updateNode exactly once
|
|
- Adapt to any node type
|
|
|
|
OPTIONAL: Call think at any point if you need to plan your approach.`;
|