aiagent.
A complete site for Claude Code + MCP agents — from setup to production-ready.
Articles — Page 2
19 postsAgent WebSocket Delegation: Auth, Envelopes, and Timeout Fallback
A practical pattern for delegating work from a server-side agent to a remote agent over WebSocket — handshake auth, typed message envelopes, and timeout fallback that doesn't leak tasks.
6 minMCP Config Scopes in Claude Code: User vs Project vs Local
How .mcp.json scope resolution works in Claude Code, where to put env vars, and a verification flow that catches broken servers before they break your session.
5 minClaude Code Slash Commands with $ARGUMENTS: A Practical Guide
How $ARGUMENTS lands inside slash command prompts, parsing patterns that survive whitespace and quoting, and validation strategies before the agent runs.
6 minA Tool Registry Pattern for pydantic-ai: Type-Safe Args, Typed Results, and Error Propagation
Build a pydantic-ai tool registry where Pydantic models define tool arguments, results return as Result<T>, and errors propagate without breaking the agent loop.
6 minBuild an MCP stdio Server in Python: JSON-RPC Framing, Tool Registration, Error Envelopes
A working Python MCP stdio server from first principles — JSON-RPC 2.0 framing on stdin/stdout, tool registration, and error envelopes that don't crash the host.
6 minAn AI Agent Pattern of the Day: Reflexive Tool-Use with Bounded Retries
A practical, code-forward pattern for AI agents that need to call tools reliably — reflexive retries with bounded backoff, structured error capture, and an explicit escape hatch.
6 minThe Claude Code Skill YAML Contract: A Field-by-Field Guide
What every field in a Claude Code skill's YAML frontmatter actually does — name, description, input shape, version semantics, and how discovery differs from invocation.
8 min