Getting Started
Find your first trusted AI agent in under five minutes.
Here's everything you need to start using Clawdbase - no account required to browse.
Browse the directory
Go to clawdbase.ai. You land directly on the directory - a searchable, filterable grid of every agent and skill in the catalog.
Each card shows: the agent name, creator handle, a short description, and two score badges if the creator is verified.
Understanding the score badges
Every verified listing shows two numbers side by side.
How trustworthy the creator's verified identity is
HighHow certain we are about that trust assessment
HighScores are color-coded by tier. A blue badge means high trust (75+), amber means medium (50–74), and red means low (below 50). Listings without badges are unverified - the creator hasn't registered yet.
Always read both scores together: a 90 trust with 30 confidence is a weaker signal than an 80 trust with 85 confidence.
Filtering and search
The toolbar above the grid has everything you need to narrow results:
Open an agent page
Click View Agent on any card. The detail page shows:
- Full description, tags, and runtime requirements
- Trust and confidence score meters
- Live GitHub stats fetched on every visit (stars, forks, open issues, last commit)
- An interactive lineage map - who created it, what it was forked from, what's been built on top
Use an agent
Each detail page has a View Source link to the GitHub repository. If the agent exposes an MCP server, you'll see a "Use via MCP" section with a ready-to-paste config for Claude Desktop, Cursor, or any MCP-compatible host.
To add Clawdbase-listed agents to Claude Desktop or Cursor, see the MCP Integration guide.