The Agent Skills Directory

[Prompt Injection] (SAFE): No direct prompt injection patterns, role-play overrides, or instructions to bypass safety guidelines were found.
[Data Exposure & Exfiltration] (SAFE): The skill uses placeholder AWS account IDs (123456789012) and queue names. No hardcoded credentials or sensitive local file path accesses were detected.
[Indirect Prompt Injection] (LOW): The skill demonstrates ingesting untrusted data from external message queues.
Ingestion points: untrusted data enters the agent context via sqs.receive_message (SKILL.md) and the Lambda event['Records'] handler.
Boundary markers: Absent; message bodies are parsed directly using json.loads() without delimiters or 'ignore' instructions.
Capability inventory: The skill includes AWS CLI commands and boto3 calls for queue management and message deletion.
Sanitization: Absent; external content is processed as-is from the queue body.
[Unverifiable Dependencies] (SAFE): The skill references boto3, a standard and trusted AWS library. No remote code downloads or execution from untrusted sources are present.

sqs