po-once
Pass
Audited by Gen Agent Trust Hub on Apr 24, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill transmits content and metadata to external domains fastidious-elephant-379.convex.site and dynamic-lapwing-647.convex.site. This activity is the primary purpose of the skill for interacting with the Po Once agent API. Evidence: The
requestWithBaseUrlfunction inscripts/po-once.cjsuses thefetchAPI to communicate with these endpoints.\n- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it ingests untrusted data from local files and user-provided strings for captions and keywords. Ingestion points: Media file content in theuploadcommand and text arguments in thecontent:createandkeyword-searchcommands. Boundary markers: Absent. Capability inventory: Bash tool execution ofscripts/po-once.cjs, which performs file reads viafs.readFileSyncand network requests viafetch. Sanitization: Sensitive fields are redacted in output, but raw input is passed to the API.
Audit Metadata