The Agent Skills Directory

[PROMPT_INJECTION]: No patterns of instruction override, safety bypass, or role-play injection were detected. The skill maintains a consistent focus on its stated purpose of document review.
[DATA_EXPOSURE_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file path access, or unauthorized network operations was found. The instructions explicitly emphasize document confidentiality as a core rule.
[OBFUSCATION]: No Base64, zero-width characters, homoglyphs, or other encoding techniques were found in the skill content.
[REMOTE_CODE_EXECUTION]: The skill does not include any package installations or remote script execution patterns.
[DYNAMIC_CONTEXT_INJECTION]: No instances of the !command syntax for shell execution at load time were found in the SKILL.md file.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data (legal documents); however, it lacks dangerous capabilities such as file system writing or network access, which limits the risk profile.
Ingestion points: User-provided legal documents (processed during review tasks).
Boundary markers: The skill instructions do not specify explicit delimiters for external content.
Capability inventory: None (the skill is restricted to text generation and analysis).
Sanitization: No explicit sanitization of document content is described.

Legal Document Review