Skip to content
Lesson 4 of 757% of the course

Every skill, connector and tool Claude used

Use the Capabilities view of Usage Insights for Claude Code to list every skill with its plugin, every MCP server with each of its tools, and every built-in tool, subagent and slash command, with calls, success and last use.

Nothing loads from YouTube until you press play. Then Google receives your IP address and may set its own storage. Narrated with an AI-generated voice. Privacy

The Me view shows the top of each list. The Capabilities view shows all of it: every skill, every MCP server down to each tool you called on it, every built-in tool, subagent and slash command, with how often, how well, and when you last used it.

Open Capabilities

Press Capabilities in the list of views. The range, the person and the project you chose still apply. The top row counts what you used: skills, connectors with their MCP tools, built-in tools, subagents, slash commands, and how many of them are new this range.

Every skill, its plugin, and who started it.

Skills and plugins

Skills lists every skill with its plugin (a skill named anthropic-skills:pdf belongs to the anthropic-skills plugin), how many times it ran, how many of those you started by name and how many Claude started on its own, in how many sessions, when it last ran, and a line of its last 30 days. A skill with no use before the range wears a new chip.

Plugins adds up each plugin's skills and commands, and under it every slash command you typed, such as /review or a plugin's own command.

Every connector, tool by tool

Connectors gives each MCP server a card: its calls, the sessions it was used in, when it was last used, and its success rate with a check (97% or better) or a warning. Inside the card, every tool you called on that server has its own row.

Which tools of each server you really use, and which ones fail.

A server whose rate drops usually has one tool at fault. The card shows which one, so you know whether to fix a token, a permission or the server itself.

Built-in tools and subagents

Tools used lists every built-in tool with its calls, success rate, rejected calls, sessions, last use and 30-day line. Subagents / research lists each agent type with its runs, average time and tool calls per run, and under it how often your hooks and scheduled tasks ran.

Habits and efficiency

Back in Me, the Habits & efficiency row says how you and Claude work together:

How much Claude does per prompt, and when you ask.
  • Tool calls per prompt and the longest run of tool calls between two prompts: how much Claude does on its own
  • Cost per session, per prompt and per commit: what an outcome costs
  • Outside working hours: prompts in the evening and at weekends
  • Busiest time and interruptions: when you work most, and how often you stopped Claude mid-answer
Your progress stays in this browser only.
Contact