Claude Gladiator FAQs

Question 1

What is Claude Gladiator?

Accepted Answer

Claude Gladiator is an MCP (Model Context Protocol) server designed to help Claude Code learn autonomously from its operational experiences. It enables continuous self-improvement by observing patterns and reflecting on actionable insights.

Question 2

Is Claude Gladiator secure and private?

Accepted Answer

Yes, privacy is a core feature. Claude Gladiator operates offline with zero external dependencies, using append-only storage. All learning and data remain local to your environment, ensuring complete privacy and speed.

Question 3

How does Claude Gladiator facilitate learning for Claude Code?

Accepted Answer

It leverages two core commands: `gladiator_observe` to record patterns, failures, or conventions, and `gladiator_reflect` to cluster these observations. Based on these clusters, it recommends updating existing rules, hooks, or skills, or creating new ones, thereby enhancing Claude's intelligence.

Question 4

What are 'rules, hooks, and skills' that Claude Gladiator updates?

Accepted Answer

These refer to the various configuration artifacts within Claude Code that define its behavior, automation, and capabilities. Claude Gladiator helps refine these artifacts (e.g., in `~/.claude/rules/`, `~/.claude/hooks/`, `~/.claude/skills/`) to make Claude more effective and aligned with your workflow.

Question 5

Can observations be automated, or must they be manual?

Accepted Answer

While you can manually record observations using `gladiator_observe`, an optional plugin provides automated observation hooks. This allows Claude Gladiator to automatically observe tool failures and prompt reflection at session-end, streamlining the learning process.

Claude Gladiator

Claude Gladiator

Key Features

Use Cases

Key Features

Use Cases