The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions found that attempt to override system prompts or bypass safety filters.
[Data Exposure & Exfiltration] (SAFE): The script does not access sensitive file paths (e.g., SSH keys, AWS credentials) or perform network operations to exfiltrate data.
[Obfuscation] (SAFE): The code is written in clear, standard Python without any use of Base64 encoding, zero-width characters, or homoglyphs.
[Unverifiable Dependencies & RCE] (SAFE): The skill only depends on the standard pandas library. There are no patterns of downloading and executing remote scripts or using dangerous functions like eval() or exec().
[Indirect Prompt Injection] (LOW):
Ingestion points: Data enters via the df (DataFrame) and data (dict) arguments in scripts/validators.py.
Boundary markers: Absent; the skill processes data directly without delimiters.
Capability inventory: None; the provided scripts perform only in-memory data transformations and validation checks.
Sanitization: Absent; the skill does not filter for natural language instructions within the data.
Assessment: While it processes untrusted data, the lack of side-effect capabilities (no file writes, network calls, or command execution) within the script logic limits the risk to the agent's internal reasoning context.

data-validation