The Agent Skills Directory

[SAFE]: The skill provides comprehensive and accurate information about using Amazon Transcribe, focusing on production heuristics and security best practices.
[COMMAND_EXECUTION]: The skill contains a Python script in SKILL.md that uses the standard boto3 library to interact with AWS services. The script uses placeholder values for bucket names and job parameters, following standard safe coding practices for documentation.
[DATA_EXPOSURE_EXFILTRATION]: No sensitive data, hardcoded credentials, or exfiltration patterns were detected. The skill explicitly advises users to avoid putting sensitive data in metadata fields like job names and tags, and to use least-privilege IAM roles.
[INDIRECT_PROMPT_INJECTION]: While the skill involves processing external data (audio for transcription), it includes multiple warnings about human review requirements for sensitive outputs like Medical or PII redaction, reducing the risk of downstream exploitation of untrusted data.
[EXTERNAL_DOWNLOADS]: The Python script requires the boto3 package, which is the official, well-known AWS SDK. No other external dependencies or remote script downloads were found.

amazon-transcribe