dub-links-api
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- COMMAND_EXECUTION (SAFE): The skill utilizes system commands
curlandjqto perform API operations. These calls are well-defined and strictly limited to the Dub.co API endpoints. - DATA_EXFILTRATION (LOW): The skill performs network operations to
api.dub.co. While this is the intended functionality, the domain is not on the pre-approved whitelist for automated network access, resulting in a LOW severity per standard policy. - CREDENTIALS_UNSAFE (SAFE): API keys are managed through environment variables (
DUB_API_KEY) and are not hardcoded in the skill definition. - INDIRECT_PROMPT_INJECTION (LOW):
- Ingestion points: API responses from
api.dub.cocontaining user-controlled link metadata (titles, descriptions). - Boundary markers: Absent; the agent is not instructed to treat API response content as untrusted.
- Capability inventory: Full CRUD capabilities on links via
curl(POST, PATCH, DELETE). - Sanitization: None; the skill parses raw JSON with
jqand presents it to the agent context without escaping or validation.
Audit Metadata