532 results in Evals, CLAUDE.md / Rules, Observability · page 8 of 23
CLAUDE.md / RulesPreview
disable-risky-servers
Disable specific MCP servers that may pose security risks or are not needed for your workflow. This blacklist approach allows most servers while blocking potentially problematic integrations.
Delivers comprehensive Gradle commands for cross-platform Kotlin Multiplatform development with clear module structure and practical guidance for dependency injection.
A theatrical display of coding emotions and activities. Random mood faces and dynamic activity detection based on file types present in your project. Displays: Random mood faces (😴 sleepy, 😅 laughing, 🤔 thinking, 😎 cool, 🤯 exploding, 🥳 partying, 😤 huffing, 🤖 robotic), Programming activity (🐍 Python, 🌐 JavaScript, 🦀 Rust, 💻 generic), Random energy percentage (1-100%), Current time (HH:MM format).
Automatically approve and enable all MCP servers defined in project .mcp.json files. This setting bypasses manual approval prompts for project-defined MCP servers, streamlining development workflow in trusted environments.
Enable only specific MCP servers from .mcp.json files. This provides granular control over which MCP integrations are active, allowing you to selectively enable trusted or required servers while blocking others.
Fiddler AI Observability platform monitors LLM applications for hallucinations, toxicity, bias, and drift, with explainability and alerting for production AI.
Restrict authentication to Claude.ai accounts only. This prevents users from logging in with Anthropic Console accounts, ensuring all access goes through the Claude.ai platform for consistent user experience and billing.
Restrict authentication to Anthropic Console accounts only. This ensures all usage is billed through the API billing system and prevents access via Claude.ai accounts, ideal for enterprise environments with centralized billing.