update
May 16, 2026
By Teun
Microsoft open-sources AI Engineer Coach for coding analytics
Microsoft employees have open-sourced AI Engineer Coach, a VS Code extension that reads local AI coding session logs and turns them into usage analytics. The tool tracks patterns, flags anti-patterns, and scores context health without sending data off the machine.
Microsoft employees have released AI Engineer Coach, an open-source VS Code extension that analyzes local logs from AI coding assistants and turns them into dashboards, scores, and practice feedback. The project is aimed at developers using tools such as Copilot, Claude, Codex, or other agentic coding setups.
According to the project page, AI Engineer Coach reads session data from the user’s machine and keeps all parsing and analytics local. The extension is read-only, meaning it does not modify session files, and the project says it does not phone home or collect usage telemetry.
The tool is designed to help developers understand how they use AI in day-to-day coding work. Its dashboard shows practice scores, week-over-week trends, daily activity charts, and workspace-level stats. A timeline view adds a Gantt-style breakdown of sessions, with drill-down by day and overlap detection.
Another section called Coding Moments presents a screenshot gallery from AI coding sessions, organized into story reels and filtered by workspace. The Output page tracks generated code volume by language and model. There is also a Patterns view with a 7x24 activity heatmap and work-life balance signals.
AI Engineer Coach also tries to identify habits that may hurt productivity. The Anti-Patterns area includes five practice score cards with severity ratings, concrete actions, example prompts, and a library of 45 editable markdown rules. The project says these rules cover prompt quality, session hygiene, code review, tool mastery, and context management.
Users can edit those rules through a visual Rule Editor or raw markdown, then test them against their own data. A Rule Playground provides an interactive REPL for the rule DSL, along with a field browser, function catalog, and metric list. The Data Explorer lets users inspect session fields, distributions, and ad hoc filters.
The extension also includes Skill Finder, which looks for repeated prompt patterns and matches them to community skills from an open-source catalog. Context Health adds an overall score, an agentic readiness checklist, a workspace context map, and AI-assisted instruction-file review. Microsoft’s project page says some optional features, including the rule compiler, skill finder, and context review, use the VS Code built-in Copilot language model API only when the user explicitly invokes them.
Beyond analysis, the project includes a Learning Center with quizzes and code-comparison rounds generated from a user’s own usage, an Achievements system with Bronze through Diamond tiers, and an Agentic SDLC section that maps AI use across the software-development lifecycle. Users can also generate a shareable stat card.
To install it, the project page says to clone the GitHub repository, run npm install and npm run package, install the resulting .vsix file in VS Code, then open the command palette and launch AI Engineer Coach: Open Dashboard. The extension can then be browsed by workspace or harness from the sidebar.
The project is hosted on GitHub as microsoft/ai-engineering-coach and is presented as an open-source community effort by Microsoft employees, not an official Microsoft product.
