The lab notebook
Field notes on AI engineering
824 articles · page 18 of 28
Hybrid Search vs Pure Vector Search for RAG: A Practical Guide
A practitioner's comparison of hybrid search vs vector search in RAG systems, with code, tradeoffs, and a decision framework.
Building an Internal Knowledge Agent for Engineering Teams
A practical guide to building an internal knowledge agent that actually answers engineering questions correctly, with retrieval, tools, and evals.
Interpreting DeepEval Score Distributions Across a Test Run
Learn how to read DeepEval score distributions across a full test run instead of chasing pass/fail counts one case at a time.
Interpreting Low Ragas Scores: A Diagnostic Checklist
A practical checklist for diagnosing why your Ragas scores are low and fixing the retrieval, generation, or dataset issue causing it.
Is a 30-Day AI Agent Bootcamp Worth It? An Honest Breakdown
A realistic look at what a 30-day AI agent bootcamp can and can't do for your career, before you spend the money.
Building a Legal Research Agent: Accuracy Over Speed
Why a legal research AI agent must trade speed for verifiable accuracy, and how to actually build one that doesn't hallucinate citations.
MCP Client Comparison: How Different AI Tools Support MCP
A practical comparison of how Claude Desktop, Claude Code, Cursor, and custom clients implement MCP, and what it means for your server design.
MCP for Local Development: Running Servers on Your Own Machine
A practical guide to running, debugging, and securing MCP servers locally before you ever deploy one to production.
MCP for Teams: Building an Internal Tool Marketplace
A practical guide to turning scattered internal scripts and APIs into a governed MCP tool marketplace your whole engineering org can discover and reuse.
MCP Prompts Explained: Reusable Prompt Templates as a Protocol Feature
MCP Prompts let servers ship reusable, parameterized prompt templates that any client can discover and run consistently.
Rate Limiting and Cost Control for Agent-Facing MCP Tools
A practical guide to rate limiting, quotas, and cost controls for MCP servers so autonomous agents can't quietly burn your budget or your APIs.
MCP Resources vs Tools vs Prompts: The Three Primitives Explained
A practical breakdown of MCP's three core primitives — Resources, Tools, and Prompts — with code examples showing when to use each.
MCP Server for Analytics: Letting Agents Query Your Data Warehouse
A practical guide to building an MCP server that lets AI agents safely query your data warehouse without hardcoding SQL into every prompt.
MCP Server for Calendar Access: Scheduling Through Agents
Learn how to build an MCP server that gives AI agents safe, structured access to calendars for real scheduling work.
MCP Server for CRM Access: Reading and Updating Customer Records
A hands-on guide to building an MCP server that lets AI agents safely read and update CRM records with proper schemas and guardrails.
MCP Server Discovery: How Clients Find and Trust Servers
A practical guide to how MCP clients discover, verify, and connect to servers safely.
MCP Server for File Systems: Safe Read/Write Access for Agents
How to build an MCP file system server that gives AI agents safe, sandboxed read/write access without risking your whole disk.
Building an MCP Server for Your Database (Postgres Example)
A practical, code-first guide to wrapping a Postgres database in an MCP server so any AI agent can query it safely.
MCP Server for Email: Reading, Drafting and Sending Safely
A practical guide to building an MCP server for email that reads, drafts, and sends without turning your inbox into an autonomous liability.
MCP Server for GitHub: Automating Issues, PRs and Reviews
A practical guide to using the MCP server for GitHub to automate issue triage, pull requests, and code reviews from your AI agent.
MCP Server for Slack: Letting Your Agent Read and Post Messages
A practical guide to building and using an MCP server for Slack so AI agents can read channels, search threads, and post messages safely.
MCP Server Health Checks: Keeping Tool Servers Reliable
A practical guide to designing MCP server health checks that catch failures before your AI agents do.
MCP Server Input Validation: Preventing Malformed Tool Calls
Learn how to validate MCP tool inputs at the schema, runtime, and semantic layers to stop malformed calls before they reach your business logic.
MCP Server Logging and Observability for Production Agents
A practical guide to structured logging, tracing, and metrics for MCP servers running real agentic workloads in production.
MCP in Python: Building Your First Server with the Python SDK
A hands-on guide to building your first Model Context Protocol server in Python, from tools and resources to running it with a real client.
MCP Server Security: Authentication, Scoping and Rate Limiting
A practical guide to authenticating, scoping, and rate limiting MCP servers before you connect them to an autonomous agent.
MCP Server Timeout Handling: Avoiding Hung Agent Sessions
A practical guide to diagnosing and fixing MCP server timeouts before they freeze your agent sessions in production.
MCP Server Versioning: Handling Breaking Changes Gracefully
A practical playbook for versioning MCP servers so tool schema changes stop breaking agents in production.
Building an MCP Server for a SaaS API You Don't Control
A practical guide to wrapping a third-party SaaS API in an MCP server when you can't change the upstream, its auth, or its rate limits.
Connecting MCP Servers to Claude Desktop: A Setup Guide
A practical, step-by-step guide to installing, configuring, and debugging Model Context Protocol servers in Claude Desktop.