Blog

Notes from the bridge between docs and agents.

Short, opinionated essays on MCP, AI tooling, and the operational reality of shipping AI-native developer products.

Editorial cover: bold serif headline 'Why better docs won't fix AI hallucinations' on a deep matte-black background, with a small Cursor terminal panel on the right highlighting one invented line labelled INVENTED.
Featured··8 min read

Why better documentation won't fix AI hallucinations

Documentation was written for humans. AI agents need infrastructure. Here is the structural problem nobody is solving — and what it looks like when you fix it.

Read the full essay
Floating glass cards labelled API Reference, OpenAPI Spec, Crawled Pages, Live Examples and MCP Server orbiting a glowing assistant orb on a dark navy background.
·6 min read

Stop pasting docs into Cursor. Let your agent borrow what others built.

The fastest engineers in 2026 don’t copy-paste API docs into their prompts. They expose a clean MCP server and let every editor pull from the same source of truth.

Read post
Glassmorphism cards for Cursor IDE, Claude Desktop, VS Code, Claude Code and Windsurf orbiting a glowing blue neural brain on a dark background.
·5 min read

One MCP server, six editors. Why the registry shape matters.

Cursor, Claude Desktop, VS Code, Windsurf, Codex and Zed all speak MCP — but they consume it differently. Here’s how doc2mcp produces a single artifact that fits all of them.

Read post
Glass cards labelled Bearer Token, RLS Policies, Signed Webhooks, Audit Logs and Zero Trust orbiting a glowing shield with a padlock on a dark navy background.
·7 min read

Tokens, headers, and the quiet leaks AI apps keep shipping

Bearer tokens in URLs, chat IDs in console logs, Authorization headers visible to every extension on the page. A pragmatic checklist for AI app security.

Read post
Glass cards labelled Diff, Auto-sync, Push, Cron and Up to date orbiting a glowing sync-icon orb on a dark navy background.
·5 min read

Your MCP should auto-sync. Here's how doc2mcp does it.

Vendor docs change quietly. Your AI agent doesn't notice until it ships a broken API call. Content hashing + cron + diff-aware regeneration is the fix.

Read post
Glass cards labelled JSON Schema, REST Routes, OpenAPI 3.1, Webhooks and v2.1.0 orbiting a glowing schema-tree sphere on a dark navy background.
·6 min read

OpenAPI is not docs. Stop shipping it to your AI agent.

Swagger specs are necessary, not sufficient. Routes don't tell a model when to call which endpoint. Workflows do.

Read post
Glass cards labelled Latency p95, Cold Starts, Token Burn, Error Rate and Traces orbiting a glowing EKG-pulse sphere on a dark navy background.
·7 min read

Observability for AI agents: what to log when the model is the bug

Latency, cold starts, token burn, error rates, traces. The five panels every team running an agent in production should be staring at by week two.

Read post
Glass cards labelled Deploy Agent, Plan Task, Browse Web, Run Tools and Talk Back orbiting a glowing terminal-cursor orb on a dark navy background.
·5 min read

Agents are just tools and loops. Stop overthinking it.

Strip the marketing. An AI agent is a model with a set of typed tools, a planner, and a stopping rule. Treat it that way and shipping gets boring (good).

Read post
Glass cards labelled System Prompt, Temperature, Chain of Thought, Tool Calls and Evals orbiting a glowing prompt-input sphere on a dark navy background.
·6 min read

Prompting still matters. Here's the cheap stuff that works.

You don't need a 4,000-token system prompt to ship a good AI feature. You need clear roles, three examples, and a stopping condition. Here's a working template.

Read post