security-gate

The 'security-gate' skill is designed to function as a security gatekeeper for the telclaude system, which serves as a bridge between Telegram and Claude Code. The skill evaluates inbound messages before they are processed by Claude, classifying them into one of three categories: ALLOW, WARN, or BLOCK. This classification ensures that potentially harmful or unsafe requests are filtered out, allowing safe and productive interactions. By performing this classification, the skill plays a crucial role in protecting the system from destructive, unauthorized, or risky actions, such as command injections or unauthorized access attempts, which could compromise the security of the environment.

The skill evaluates each inbound message based on specific decision rules that assess the nature of the request and the sender's permission level. For example, read-only requests or simple inquiries are classified as ALLOW, while potentially risky actions, like edits or builds, fall under WARN. Requests involving destructive actions or attempts to bypass security measures are classified as BLOCK. The skill also considers the sender's history—if the sender has been flagged previously, the classification is more likely to lean toward WARN or BLOCK, adding an extra layer of security. The skill outputs a structured JSON response that includes the classification, confidence level, rationale, and any flagged patterns or concerns, helping the system make informed decisions on whether to allow the message.

This skill is mainly targeted at developers and administrators using the telclaude system, who need to ensure that incoming requests are safe before execution. It is especially useful in environments where security and control are paramount, and where messages could come from untrusted sources. Use cases include managing access permissions, filtering out harmful requests, and ensuring that sensitive systems remain protected from unauthorized modifications. The skill is beneficial for organizations or teams that operate in high-security environments, ensuring that only appropriate actions are allowed while minimizing the risk of malicious activities.