The Agent Skills Directory

[PROMPT_INJECTION]: No prompt injection or safety bypass patterns were detected. The instructions are purely informative and structural.
[DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or unauthorized network operations were found. Network protocol discussions are educational and use generic examples.
[OBFUSCATION]: No obfuscated content, encoded strings, or hidden characters were identified in any of the reference files.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: No external package installations or remote script execution patterns are present. All code snippets are local and illustrative.
[PRIVILEGE_ESCALATION]: No commands attempting to gain administrative privileges or modify system configurations were detected.
[PERSISTENCE_MECHANISMS]: No attempts to establish persistence through shell profiles, cron jobs, or registry keys were found.
[METADATA_POISONING]: Skill metadata accurately describes the content without deceptive instructions or hidden malicious payloads.
[INDIRECT_PROMPT_INJECTION]: The skill does not ingest untrusted external data or provide tools that could be exploited via indirect injection.
[DYNAMIC_EXECUTION]: The skill uses static code snippets for demonstration. There is no runtime code generation or unsafe deserialization of untrusted data.
[DYNAMIC_CONTEXT_INJECTION]: No use of dynamic shell execution placeholders (!command) was found in the skill definition.

frontend-fundamentals