Agent skills for building AI applications with anchor — the context engineering toolkit for Python.
What Are Skills?
Skills are structured knowledge files that teach AI coding agents (like Claude Code) how to use libraries correctly. They provide:
- Correct API signatures and import paths
- WRONG/CORRECT patterns to prevent common mistakes
- Working code examples for every major feature
- Parameter reference tables
Available Skills
| Skill | Description |
|---|---|
anchor-fundamentals | Core pipeline, context items, formatters, observability |
anchor-rag | Retrieval, chunking, ingestion, reranking, evaluation |
anchor-memory | Conversation memory, persistence, eviction, graph memory |
anchor-agents | Agent framework, tools, skills, MCP integration |
anchor-llm-providers | LLM providers, streaming, fallback, cost tracking |
Installation
git clone https://github.com/Astrointelligence-dev/astro-skills.git cd astro-skills ./install.sh
Then restart Claude Code. The skills will activate automatically when you work with anchor code.
Manual Installation
Copy the .claude-plugin/ and config/ directories to ~/.claude/plugins/astro-skills/.
Usage
Once installed, the skills trigger automatically based on context:
- Writing a
ContextPipeline? →anchor-fundamentalsactivates - Setting up retrieval? →
anchor-ragactivates - Adding conversation memory? →
anchor-memoryactivates - Building an agent with tools? →
anchor-agentsactivates - Configuring LLM providers? →
anchor-llm-providersactivates
License
MIT