The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple external cloud providers and local synchronization artifacts, creating a potential surface for indirect prompt injection.\n
Ingestion points: The skill retrieves files, metadata, and logs through GoogleDriveForensicAcquisition.list_all_files (SKILL.md), acquire_s3_objects (scripts/agent.py), and local file discovery (scripts/process.py).\n
Boundary markers: The provided implementation lacks specific boundary markers or instructions to the agent to disregard potentially malicious commands embedded in the evidence data.\n
Capability inventory: The skill utilizes network capabilities for API interaction, extensive file system access for writing evidence, and database analysis tools.\n
Sanitization: Content retrieved from cloud sources and local databases is not sanitized or validated before being logged or processed.\n- [EXTERNAL_DOWNLOADS]: The skill documentation recommends installing several official libraries from well-known technology providers to enable API connectivity.\n
Evidence: Mentions google-api-python-client, msal, boto3, and dropbox in SKILL.md and references/api-reference.md.\n- [COMMAND_EXECUTION]: Instructions are included for running a specialized forensic tool via the command line for endpoint artifact collection.\n
Evidence: A specific command line for kape.exe is provided in SKILL.md to target cloud storage artifacts on a local system.

performing-cloud-storage-forensic-acquisition