Claude Gladiator MCP

Example Use Cases

• 1

  Automatically observe and learn from tool failures and user corrections.

• 2

  Reflect on accumulated observations to improve rules, hooks, and skills.

• 3

  That require offline, dependency-free learning loops integrated with Claude Code.

Description

Claude Gladiator MCP provides a learning loop for Claude Code by recording observations of tool failures, user corrections, and codebase patterns, then clustering and analyzing them to suggest updates or new artifacts. It operates offline with zero dependencies and uses clustering and IDF-weighted scoring algorithms to improve Claude's performance over time. The MCP server integrates with skills and plugins to automate observation and reflection processes, enhancing discoverability and proactive learning.