The Agent Skills Directory

[PROMPT_INJECTION]: No instructions attempting to override agent behavior, bypass safety protocols, or extract system prompts were detected across the documentation.
[DATA_EXFILTRATION]: No patterns indicative of sensitive data exposure or exfiltration were found. References to credentials use standard placeholders or environment variable practices. Network operations in examples target well-known services or local endpoints.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading established monitoring tools (Prometheus, OpenTelemetry Collector, Jaeger) from their official GitHub repositories and trusted domains. These references are documented neutrally and follow safe practices.
[COMMAND_EXECUTION]: Shell command examples are limited to standard system administration tasks (e.g., systemctl, package management, log rotation) necessary for the deployment of observability infrastructure.
[OBFUSCATION]: Analysis of all files confirmed the absence of Base64 encoding, zero-width characters, homoglyphs, or other obfuscation techniques used to hide malicious content.
[DYNAMIC_EXECUTION]: Python and Node.js code snippets provide benign examples for log parsing and metric collection. No unsafe dynamic execution patterns or runtime compilation of untrusted code were identified.
[PERSISTENCE_MECHANISMS]: While the guide mentions cron for log rotation and scheduled analysis, these are provided as legitimate operational examples and do not represent malicious persistence attempts.

implementing-observability