Microsoft open-sourced AI Engineer Coach, a VS Code extension that analyzes how developers use AI coding agents. The tool reads local logs from GitHub Copilot, Claude Code, Codex CLI, OpenCode, and Xcode into a single dashboard.
Key facts
- Analyzes sessions from 5 AI coding tools.
- Scores across 5 workflow quality categories.
- Includes 45 anti-pattern detection rules.
- Uses markdown-based rule engine with expression language.
- MIT licensed, fully open-source, zero telemetry.
Microsoft released AI Engineer Coach as an open-source VS Code extension (also compatible with Cursor and Antigravity) that treats developer-AI interaction like an observability tool treats production systems. [According to @akshay_pachaar] The extension reads local session logs from five major AI coding tools — GitHub Copilot, Claude Code, Codex CLI, OpenCode, and Xcode — and presents a unified dashboard.
How It Scores Developers
The tool scores workflow across five categories: prompt quality, session hygiene, code review, tool mastery, and context management. It ships with 45 anti-pattern detection rules covering issues like prompts lacking file context, mega sessions that drift off-topic, auto-approving terminal commands without a devcontainer, and burning premium tokens on trivial questions. Each finding includes what went wrong, how to fix it, and a real example from the user's own sessions.
The Rule Engine Is the Differentiator
The rule engine is the standout feature. Every detector is a markdown file with a small expression language, letting users tune thresholds, write new rules, or describe one in plain English and let Copilot scaffold it. There's also a Skill Finder that spots repeated prompt patterns and turns them into reusable skills. Everything runs locally, is read-only by design, and sends zero telemetry.
The Unique Take
The AP wire would frame this as another dev tool release. The structural observation is different: after two years of making AI agents faster, the industry has almost no tooling to measure how effectively developers work with them. AI Engineer Coach fills that gap by applying observability principles — logging, scoring, alerting — to the developer-AI interaction layer. It's a sign that the next frontier isn't agent speed but agent workflow quality, and Microsoft is betting open-source community contributions will define the anti-pattern library faster than any internal team could.
What to watch
Watch for community adoption velocity — how many custom anti-pattern rules get contributed in the first 90 days. Also monitor whether JetBrains or Cursor builds a competing version, signaling that workflow observability becomes a standard feature, not a niche extension.
