people-strategy（人才战略）

People-Strategy Agent Skill

Overview

A people relationship management agent skill that provides persistent graph-based storage using SQLite database. This skill enables AI agents to create, manage, and query a network of people and their relationships, making it ideal for building personal CRMs, organizational charts, mentorship networks, and collaboration mapping.

Capabilities

Person Management

• Create People: Add new people with comprehensive information (name, role, relation, organization, character, notes)
• Read People: Retrieve individual people or search/filter across the database
• Update People: Modify any person's information
• Delete People: Remove people from the database (cascades to relationships)
• Search: Full-text search across names, roles, and organizations
• Filter: Filter by organization or relation type

Relationship Management

• Create Edges: Establish directed relationships between people
• Query Relationships: Get incoming, outgoing, or all relationships for a person
• Update Edges: Modify relationship types and descriptions
• Delete Edges: Remove specific relationships
• Bidirectional Support: Track both directions of relationships

Graph Operations

• Full Graph Export: Export entire network as JSON (nodes + edges)
• Person Network: Get complete network view for an individual
• Relationship Types: Support for custom relationship types (reports_to, mentors, works_with, etc.)

Database Schema

Tables

people

• id: INTEGER PRIMARY KEY
• name: TEXT NOT NULL
• role: TEXT
• relation_to_me: TEXT
• organization: TEXT
• character: TEXT
• notes: TEXT
• created_at: TIMESTAMP
• updated_at: TIMESTAMP

edges

• id: INTEGER PRIMARY KEY
• from_person_id: INTEGER (FK to people.id)
• to_person_id: INTEGER (FK to people.id)
• relationship_type: TEXT NOT NULL
• description: TEXT
• created_at: TIMESTAMP
• updated_at: TIMESTAMP
• UNIQUE constraint on (from_person_id, to_person_id, relationship_type)

Indexes

• idx_people_name: Index on people.name
• idx_edges_from: Index on edges.from_person_id
• idx_edges_to: Index on edges.to_person_id

Usage Examples

Command Line Interface

# Person operations
python people_skill.py add-person "Jane Doe" --role "CTO" --org "StartupXYZ" --relation "Mentor"
python people_skill.py get-person 1
python people_skill.py search "engineer"
python people_skill.py list-people --org "StartupXYZ"
python people_skill.py update-person 1 --notes "Coffee meeting scheduled"
python people_skill.py delete-person 1

# Relationship operations
python people_skill.py add-relationship 1 2 "mentors" --description "Tech career guidance"
python people_skill.py get-relationships 1
python people_skill.py list-relationships
python people_skill.py update-relationship 1 --type "coaches"
python people_skill.py delete-relationship 1

# Graph operations
python people_skill.py get-graph
python people_skill.py get-network 1

Python API

from database import PeopleDatabase
from people_skill import PeopleAgent

# Initialize
agent = PeopleAgent("people.db")

# Add people
result = agent.add_person(
    name="Alice Johnson",
    role="VP Engineering",
    relation_to_me="Former colleague",
    organization="TechCorp",
    character="Strategic thinker, empathetic leader",
    notes="Worked together 2020-2022"
)
person_id = result["person_id"]

# Add relationship
agent.add_relationship(
    from_person_id=1,
    to_person_id=2,
    relationship_type="manages",
    description="Direct manager relationship"
)

# Query network
network = agent.get_person_network(person_id)
print(f"Found {network['connections_count']} connections")

# Export graph
graph = agent.get_graph()
print(f"Graph: {graph['nodes_count']} nodes, {graph['edges_count']} edges")

Common Relationship Types

• reports_to: Hierarchical reporting (subordinate → manager)
• manages: Hierarchical management (manager → subordinate)
• works_with: Peer collaboration
• mentors: Mentorship (mentor → mentee)
• mentored_by: Inverse mentorship (mentee → mentor)
• friends_with: Personal friendship
• knows: Acquaintance
• collaborates_with: Project collaboration
• introduced_by: Connection source

Use Cases

1. Personal CRM

Track professional contacts, their roles, organizations, and how you know them. Add notes about conversations, meetings, or follow-ups.

python people_skill.py add-person "Sarah Chen" \
  --role "Product Director" \
  --org "InnovateCo" \
  --relation "Met at conference" \
  --notes "Interested in API collaboration, follow up Q2"

2. Organizational Mapping

Build and maintain organizational charts with reporting relationships.

# Add team members
python people_skill.py add-person "Mike Lead" --role "Team Lead" --org "Engineering"
python people_skill.py add-person "Dev One" --role "Engineer" --org "Engineering"
python people_skill.py add-person "Dev Two" --role "Engineer" --org "Engineering"

# Create reporting structure
python people_skill.py add-relationship 2 1 "reports_to"
python people_skill.py add-relationship 3 1 "reports_to"

3. Mentorship Network

Track mentorship relationships and career guidance connections.

python people_skill.py add-relationship 5 3 "mentors" \
  --description "Career guidance in ML/AI"

4. Collaboration Tracking

Map who works with whom on projects.

python people_skill.py add-relationship 1 2 "works_with" \
  --description "Project Phoenix collaboration"

5. Network Analysis

Analyze your professional network to identify key connections.

# View someone's complete network
python people_skill.py get-network 1

# Export for visualization
python people_skill.py get-graph > network.json

Advanced Features

Bidirectional Relationship Queries

The system automatically handles both directions:

# Get all relationships for a person
edges = db.get_all_edges_for_person(person_id)
# Returns: {"outgoing": [...], "incoming": [...]}

Cascade Deletion

Deleting a person automatically removes all associated relationships.

Duplicate Prevention

The UNIQUE constraint prevents duplicate relationships between the same two people with the same type.

Search Flexibility

Search works across multiple fields:

python people_skill.py search "engineer"
# Matches: names, roles, organizations

Integration Examples

Export to Visualization Tools

# Export graph for tools like D3.js, Cytoscape, Gephi
python people_skill.py get-graph > graph.json

Import from CSV

import csv
from people_skill import PeopleAgent

agent = PeopleAgent()
with open('contacts.csv', 'r') as f:
    reader = csv.DictReader(f)
    for row in reader:
        agent.add_person(
            name=row['name'],
            role=row['role'],
            organization=row['org'],
            relation_to_me=row['relation']
        )

Best Practices

1. Consistent Naming: Use consistent naming conventions for relationship types
2. Rich Notes: Add contextual notes to remember important details
3. Regular Updates: Keep information current with update commands
4. Character Tracking: Use the character field to remember personality traits
5. Relationship Descriptions: Add context to relationships for clarity
6. Search First: Before adding, search to avoid duplicates

API Reference

See people_skill.py for complete API documentation.

Key Classes

• PeopleDatabase: Low-level database operations
• PeopleAgent: High-level agent interface with result dictionaries

Return Format

All agent methods return dictionaries with:

• success: Boolean indicating success/failure
• message: Human-readable status message (on errors)
• Data fields specific to the operation

Error Handling

The skill includes comprehensive error handling:

• Foreign key constraints prevent invalid relationships
• Unique constraints prevent duplicate relationships
• Cascade deletion maintains referential integrity
• Try-catch blocks in CLI provide user-friendly errors

Performance

• Indexed queries for fast lookups
• Efficient graph traversal
• Suitable for networks of thousands of people
• SQLite handles concurrent reads

Future Enhancements

Potential extensions:

• Graph visualization generation
• Shortest path finding between people
• Influence/centrality scoring
• Duplicate person detection
• Export to various formats (GraphML, DOT)
• Import from LinkedIn, contact lists
• Time-based relationship tracking
• Strength/confidence scoring for relationships

License

MIT