skills/openai/plugins/mgnify-skill/Gen Agent Trust Hub

mgnify-skill

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Flexible Network Interaction]: The scripts/rest_request.py script functions as a generic REST client, accepting a target URL, method, and headers via input. While documentation points to the MGnify service, the underlying script does not restrict the target domain, which is a consideration for managing network interaction surfaces.
  • [File System Write Capability]: The skill provides parameters to save API response data to the local filesystem. It can create directories and write files to paths specified by the agent, which is a noteworthy capability for handling external data.
  • [Indirect Prompt Injection Surface]: The skill processes data from external endpoints and returns it to the agent. This represents a potential surface where instructions embedded in external content could influence the agent's behavior. (1) Ingestion points: External API data is fetched via the requests library in scripts/rest_request.py. (2) Boundary markers: The skill does not employ explicit delimiters or warnings to ignore instructions in the tool output. (3) Capability inventory: The script can perform network requests and write to arbitrary file paths. (4) Sanitization: A compaction process limits the length of returned strings, but no explicit filtering or escaping of content is performed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:20 PM
Security Audit — agent-trust-hub — mgnify-skill