From learn
Researches any topic via web, docs, and Context7, then creates a permanent skill for Claude on code libraries, APIs, concepts, tools, or domains. Invoke via /learn <topic>.
npx claudepluginhub aviz85/claude-skills-library --plugin learnThis skill uses the workspace's default tool permissions.
Research and learn any topic, then retain the knowledge as a permanent skill. Works for:
Creates isolated Git worktrees for feature branches with prioritized directory selection, gitignore safety checks, auto project setup for Node/Python/Rust/Go, and baseline verification.
Executes implementation plans in current session by dispatching fresh subagents per independent task, with two-stage reviews: spec compliance then code quality.
Dispatches parallel agents to independently tackle 2+ tasks like separate test failures or subsystems without shared state or dependencies.
Research and learn any topic, then retain the knowledge as a permanent skill. Works for:
After learning, Claude becomes an expert on that topic in all future conversations.
/learn <topic> # Create new skill from docs
/learn <topic> --update # Update existing skill with latest docs
/learn <topic> --project # Save to project (.claude/skills/) instead of global
For code libraries/APIs - try Context7 first:
mcp__plugin_context7_context7__resolve-library-id
query: "user's task/question"
libraryName: "<topic>"
For any topic (or if Context7 doesn't have it) - use WebSearch:
WebSearch: "<topic>" guide tutorial best practices
WebSearch: "<topic>" official documentation
WebSearch: "<topic>" examples patterns
For domain knowledge - broader research:
WebSearch: "<topic>" comprehensive guide
WebSearch: "<topic>" key concepts terminology
WebSearch: "<topic>" common mistakes pitfalls
Use WebFetch to read the most relevant pages found.
Query the documentation for key information:
mcp__plugin_context7_context7__query-docs
libraryId: "<resolved-id>"
query: "getting started setup authentication common patterns"
Extract:
If docs are large, make multiple targeted queries:
Before creating the skill, use the claude-code-guide agent to verify current skill conventions:
Task tool with subagent_type: claude-code-guide
prompt: "What are the current conventions for writing Claude Code skills?
I need to know: required/optional frontmatter fields,
description best practices, and recommended structure."
This ensures the generated skill follows the latest Claude Code patterns.
Create skill at appropriate location:
~/.claude/skills/<topic>/skill.md--project: .claude/skills/<topic>/skill.mdSkill template:
---
name: <topic>
description: <What the library does>. Use when <specific triggers>. Covers: <main capabilities>.
---
# <Topic> Quick Reference
## Setup
[Installation commands and initial config]
## Core Concepts
[Key abstractions - 2-3 sentences each]
## Common Patterns
### [Pattern 1 Name]
[Code example with brief explanation]
### [Pattern 2 Name]
[Code example with brief explanation]
## API Quick Reference
| Method | Purpose | Example |
|--------|---------|---------|
| ... | ... | ... |
## Gotchas
- [Non-obvious issue 1]
- [Non-obvious issue 2]
## Links
- [Official Docs](url)
- [API Reference](url)
After creating the skill:
Do:
Don't:
When updating an existing skill:
User: /learn stripe-payments
Claude:
1. Resolves "stripe" via Context7 → /stripe/stripe-node
2. Queries: setup, authentication, common patterns, webhooks
3. Consults claude-code-guide for current skill conventions
4. Creates ~/.claude/skills/stripe-payments/skill.md with:
- API key setup
- Payment intent flow
- Webhook verification
- Error handling patterns
5. Reports: "Created stripe-payments skill. Triggers on: 'Stripe payment', 'charge card', 'payment intent'"