Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous
curlcommands for the agent to execute, enabling interaction with the Instagram Graph API for media retrieval, hashtag searching, and content publishing. - [DATA_EXFILTRATION]: The skill handles an authentication token (
INSTAGRAM_TOKEN) to communicate with Meta's official Graph API atgraph.facebook.com. This operation is the intended functionality of the integration and targets a well-known service. - [PROMPT_INJECTION]: The skill processes data from external sources (Instagram captions and media metadata), creating a surface for potential indirect prompt injection.
- Ingestion points: Media captions and metadata retrieved from the
graph.facebook.comAPI endpoints. - Boundary markers: Absent. The instructions do not specify the use of delimiters to isolate external data from the agent's internal instruction context.
- Capability inventory: Network operations via
curland local file writing to/tmp/request.jsonfor API payload preparation. - Sanitization: Absent. The skill does not provide instructions for validating or sanitizing retrieved content before the agent processes it.
Audit Metadata