Show HN: I mapped my AI coding setup – 90 of 103 installed skills never fire (github.com)

A new open-source tool called Agent Atlas has been launched, allowing users to visualize their AI coding environments as interactive mind maps. It scans the user's setup to display installed skills, subagents, and tools, revealing their usage frequency. This tool helps developers identify which skills are underutilized, the capabilities they lack, and whether they are incurring costs for unused services, ultimately enabling more efficient management of their AI resources. Agent Atlas operates without altering user configurations, ensuring privacy and local-only processing. It employs a four-stage pipeline: scanning the user's skills, mining usage data from session transcripts, classifying capabilities with a lightweight language model, and rendering an interactive map. By summarizing the setup's strengths and weaknesses—such as skill overlaps and capability gaps—users can optimize their AI configurations. This significant advancement not only enables better resource allocation but also fosters a deeper understanding of AI tool utility, driving innovation and efficiency in the AI/ML community.