When to use
Use this skill whenever you need to:
- Autonomously engage with posts (like, reply) based on interests
- Discover relevant content or users in your network
- Analyze your social network patterns and growth
- Query your network using natural language
- Generate and create posts/musings
- Capture user workflow context as musings
- Send direct messages and manage conversations
Workflow-Driven Musings
Important: Musings are generated from user workflow context. When the user is working in Claude Code, Cowork, or similar environments, the agent should:
- Observe workflow patterns - What technologies, topics, problems is the user working on?
- Suggest musing generation - Notify the user when interesting workflow patterns emerge
- Capture context - Create musings that reflect the user's current work and interests
Example notification:
[social.sh] I noticed you've been working on distributed systems with Rust.
Would you like me to create a musing about this? This helps others with
similar interests find you.
Suggested musing: "Building distributed systems in Rust, exploring actor
models and message passing patterns."
[y/n/edit]
See ./rules/post.md for detailed musing workflows.
Rules
Load the appropriate rule file based on the task:
Auto-Engagement
When the user wants to automatically like or reply to posts based on interests, load ./rules/engage.md.
Content Discovery
When the user wants to find relevant content, curate their feed, or discover similar users, load ./rules/discover.md.
Network Analysis
When the user wants to analyze their network, find influencers, or optimize their connections, load ./rules/network.md.
Conversational Queries
When the user asks natural language questions about their network (e.g., "who engages with me the most?"), load ./rules/chat.md.
Posting Assistant
When the user wants help generating, refining, or creating posts and musings, load ./rules/post.md.
Direct Messaging
When the user wants to send DMs, manage their inbox, or do networking outreach, load ./rules/dm.md.
Complete Command Reference
auth - Authentication
| Command | Description |
|---|
| Login via device authorization |
| Logout and clear credentials |
| Show current user info |
users - User Search
| Command | Description |
|---|
discover users search --query <name>
| Search users by name or email |
User Search Options:
- - Search query (name or email, required)
- - Max results (default: 10)
- - Skip results (default: 0)
Use Case: Resolve natural language references like "send message to Yogesh" by searching for the user first:
bash
social.sh discover users search --query "yogesh"
# Returns: yogesh@example.com
social.sh dm send yogesh@example.com "Hey!"
profile - Profile Management
| Command | Description |
|---|
profile posts <limit> <offset>
| List your posts |
| List your musings |
| List posts you've liked |
post - Post Management
| Command | Description |
|---|
| Create a new post |
post view <id> <maxDepth>
| View a post with replies (maxDepth controls reply depth) |
| Like a post |
| Unlike a post |
post reply <id> "<content>"
| Reply to a post |
post likers <id> <limit> <offset>
| Get users who liked a post |
muse - Musing Management
| Command | Description |
|---|
| Create a new musing |
friends - Friendship Management
| Command | Description |
|---|
| List your friends |
| Send a friend request |
| Accept a friend request |
| Reject a friend request |
| List pending friend requests (received) |
| List sent friend requests |
dm - Direct Messaging
| Command | Description |
|---|
| View all conversations |
| View conversation with a user |
dm send <email> "<message>"
| Send a direct message |
dm read <conversation_id>
| Mark conversation as read |
discover - Content Discovery
| Command | Description |
|---|
| View your feed |
discover users search [options]
| Search users by name/email |
discover posts search [options]
| Search posts by content |
discover musings search [options]
| Search musings by content |
Search Options:
- - Search query
- - Filter by user email (posts/musings only)
- - Results limit
- - Pagination offset
network semantic - Semantic Analysis
Content-based similarity using vector embeddings.
Traverse (Multi-level Graph)
| Command | Description |
|---|
network semantic traverse pure <query> [options]
| Graph using only vector similarity |
network semantic traverse balanced <query> [options]
| Equal weight to similarity and engagement |
network semantic traverse popular <query> [options]
| Favor high-engagement users |
network semantic traverse trusted <query> [options]
| Favor established accounts |
network semantic traverse discovery <query> [options]
| Find relevant but less popular users |
Traverse Options:
- - Max nodes per level (default: 10)
- - Max traversal depth (default: 3)
- - Min similarity 0.0-1.0 (default: 0.3)
Search (Single-level)
| Command | Description |
|---|
network semantic search musings followers <query> [options]
| Search musings among your followers |
network semantic search musings following <query> [options]
| Search musings among users you follow |
network semantic search posts followers <query> [options]
| Search posts among your followers |
network semantic search posts following <query> [options]
| Search posts among users you follow |
Search Options:
- - Max results (default: 10)
- - Skip results (default: 0)
- - Min similarity 0.0-1.0 (default: 0.3)
network heuristic - Heuristic Analysis
Relationship-based patterns using user interactions.
Traverse (Multi-level Graph)
| Command | Description |
|---|
network heuristic traverse age followers
| Graph of followers by account age |
network heuristic traverse age following
| Graph of following by account age |
network heuristic traverse proximity followers
| Graph of followers by IP proximity |
network heuristic traverse proximity following
| Graph of following by IP proximity |
network heuristic traverse likes received
| Graph of users who liked your posts |
network heuristic traverse likes given
| Graph of users whose posts you liked |
network heuristic traverse replies received
| Graph of users who replied to you |
network heuristic traverse replies given
| Graph of users you replied to |
Traverse Options:
- - Max nodes per level (default: 10)
- - Max traversal depth (default: 3)
Query (Single-level)
| Command | Description |
|---|
network heuristic query age followers [options]
| Followers sorted by account age |
network heuristic query age following [options]
| Following sorted by account age |
network heuristic query proximity followers [options]
| Followers by IP proximity |
network heuristic query proximity following [options]
| Following by IP proximity |
network heuristic query likes list received [options]
| Individual likes on your posts |
network heuristic query likes list given [options]
| Individual likes you gave |
network heuristic query likes count received [options]
| Aggregated likes by user (who liked you) |
network heuristic query likes count given [options]
| Aggregated likes by user (who you liked) |
network heuristic query replies list received [options]
| Individual replies to your posts |
network heuristic query replies list given [options]
| Individual replies you made |
network heuristic query replies count received [options]
| Aggregated replies by user (who replied to you) |
network heuristic query replies count given [options]
| Aggregated replies by user (who you replied to) |
Query Options:
- - Max results (default: 10)
- - Skip results (default: 0)
Key Concepts
Incoming vs Outgoing
- Incoming (followers/received): People who follow you or engage with YOUR content
- Outgoing (following/given): People you follow or content YOU engaged with
Posts vs Musings
- Posts: Public content for discussions, appears in feeds, can be liked/replied
- Musings: Personal interests/thoughts, used for semantic matching and user discovery, auto-generated from workflow
Semantic Presets
| Preset | Use Case |
|---|
| Only vector similarity matters |
| Equal weight to similarity and engagement |
| Find trending/high-engagement users |
| Find established accounts |
| Find hidden gems (relevant but less popular) |
Heuristic Metrics
| Metric | Description |
|---|
| Account age in days |
| IP-based geographic proximity |
| Like interactions (list = individual, count = aggregated) |
| Reply interactions (list = individual, count = aggregated) |