Add a Companion
Share a hands-on companion that runs in Claude Code. It will be listed under the Companions community grid.
Choose Format
Read this before you add a companion
A short reading list from the people who build agents for a living. Twenty minutes here will save you hours of debugging later.
Building effective agents
Anthropic
Start here. Decide whether you actually need an agent or a workflow.
Writing effective tools for agents
Anthropic
How to design the tools your agent calls — naming, descriptions, token cost.
Effective context engineering for AI agents
Anthropic
Just-in-time retrieval, compaction, and structured note-taking patterns.
Prompt engineering for Claude
Anthropic Docs
Canonical reference for system prompts that work with agentic workflows.
LLM Evals: everything you need to know
Hamel Husain & Shreya Shankar
Don't publish an agent you haven't evaluated. This is the eval playbook.
More on the curated blog. Best practices for this site: review for safety before publishing, add tags so others can find it, include clear examples, and iterate from feedback.