Building with AI coding agents: tools and integrations
Model Context Protocol
Using the Model Context Protocol with agents
- What Is the Model Context Protocol and When to Use It
- Building an Internal MCP Server for Your Team's Tools
- How to Audit an MCP Server Before You Connect It
- Rate Limits and Quotas for Internal MCP Servers
- Structuring MCP Tool Outputs So Agents Don't Guess
- A Rate and Concurrency Cap for Each MCP Server You Connect
- An MCP Server Health Endpoint: What Agents and Operators Need
- Deciding Between a Vendor MCP Server and One You Host
- All of Model Context Protocol
Agent tool integration
Connecting AI agents to tools and APIs
- CLI, MCP Server or Code Sandbox: Picking a Wrapper
- Exit Codes and Output Contracts for Agent Tools
- MCP vs Function Calling: Deciding What to Connect and How
- Naming Tools So a Model Picks the Right One
- Wrapping Old Code Instead of Rewriting It
- Writing Tool Documentation a Model Reads Correctly
- Backpressure for Shared Services Behind an Agent Wrapper
- Connection Pool Limits for Agent Fleets
- All of Agent tool integration
Tool schema versioning
Versioning and evolving AI tool schemas
- A Compatibility Test for Tool Schemas Before Agents Switch
- Contract Tests for Tool Schemas When a Field Changes
- Cutting Live Agents to a New Tool Schema
- Routing Tool Calls by Version When Two Schemas Run Live
- Schema Changes That Break Retry Logic: A Compatibility Suite
- When a Tool Description and Its Schema Disagree
- Writing Tool Schemas That Survive Model Upgrades
- A Registry for Tool Schemas: Versions, Hashes and Rollback
- All of Tool schema versioning
Tool call testing
Testing and monitoring AI tool calls
- Contract Tests Between Your App and an Inference API
- Replaying Recorded Error Payloads Against Every Model
- Testing a Retry Counter with Fixed Attempt Sequences
- Testing a Tool Contract with Recorded Runs
- A Canary for Model Upgrades: Shadow Calls on Real Tool Traffic
- A Thin Client That Replays One Tool Call Outside the Agent
- Asserting on the Wrapper Log, Not Just the Return Value
- Comparing Wrong-Type Arguments Across the Models You Route To
- All of Tool call testing
Agent security