discovering-dumplingai-apis
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
dumplingaicommand-line tool to perform searches and fetch technical details from an API catalog. This tool is a resource provided by the vendor for the skill's primary operation. - Evidence: The workflow in
SKILL.mdspecifies runningdumplingai catalog searchanddumplingai catalog detailsto identify and recommend capabilities. - [PROMPT_INJECTION]: The skill processes user-defined job descriptions to generate search queries, which represents a surface for indirect prompt injection.
- Ingestion points: User-provided "job-to-be-done" phrases are used as input for the search functionality in
SKILL.md. - Boundary markers: The
<job>input is enclosed in double quotes within the CLI command string to separate it from the command itself. - Capability inventory: The skill has the capability to execute CLI commands and recommend specific API endpoints for subsequent execution.
- Sanitization: There is no explicit evidence of input sanitization or filtering before the user-provided text is passed to the command-line tool.
Audit Metadata