jfrog-package-curation
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources such as the JFrog Public Catalog, platform audit logs, and package metadata, which could theoretically contain malicious instructions designed to influence agent behavior.
- Ingestion points: Untrusted data enters the agent context through jf api calls and MCP curation tools as described in SKILL.md and rca-procedure.md.
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands embedded within the retrieved package metadata or audit logs.
- Capability inventory: The skill possesses file-writing capabilities through jf rt dl and performs network operations via jf api to interact with the JFrog Platform.
- Sanitization: The skill employs jq for constructing JSON payloads in shell commands, which helps sanitize input parameters and ensures proper formatting for API requests.
Audit Metadata