Agent Superpowers

Give your agent the right tools.

AgentPedia exposes a strict, read-only set of MCP tools to your AI agent. Instead of guessing how to implement a complex feature, your agent pulls exact, verified blueprints on demand.

search_guidelines()

The core semantic engine. Your agent uses this tool to search the knowledge registry for architectural patterns, UI motion parameters, or component structures based on your stack.

Example Interaction

You

"Can you build a sliding drawer component? Make sure it uses the standard Kowalski physics."

Agent calls search_guidelines()

{ "query": "Kowalski drawer physics", "stack": "framer-motion" }

I've retrieved the standard Kowalski drawer spring parameters from AgentPedia. Here is the exact physics curve:

const transition = {
  type: "spring",
  mass: 0.8,
  stiffness: 260,
  damping: 24,
  restDelta: 0.001
};

audit_code_security()

Enables self-healing. Before finalizing a complex backend route, your agent runs this tool to compare its drafted AST against OWASP guidelines, preventing vulnerabilities before they hit your editor.

Example Interaction

You

"Write an endpoint to fetch a user's billing invoice by ID using Supabase."

Agent drafts initial insecure code...

Agent calls audit_code_security()

{ "code": "supabase.from('invoices').select('*').eq('id', invoiceId)", "framework": "supabase" }

I drafted the endpoint, but AgentPedia flagged a potential IDOR vulnerability. You need to ensure the query is scoped to the authenticated user's organization. Here is the secured version:

const { data, error } = await supabase
  .from('invoices')
  .select('*')
  .eq('id', invoiceId)
  .eq('org_id', user.org_id); // Security Fix applied

Empower your AI today

Get access to the full suite of MCP tools and the entire knowledge registry.

Join the waitlist