skills/leonardo-picciani/dataforseo-agent-skills/dataforseo-content-generation-api/Gen Agent Trust Hub
dataforseo-content-generation-api
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- Metadata Poisoning (MEDIUM): The skill metadata (SKILL.md) claims the project was generated with 'OpenAI GPT-5.2', a version that does not currently exist. This is deceptive and may be intended to mislead users or auditors regarding the skill's origin, safety, or capabilities.
- Indirect Prompt Injection (LOW): The skill is designed to process untrusted external text for tasks such as paraphrasing and summarization but lacks explicit boundary markers or instructions to prevent the agent from executing commands embedded in that text.
- Ingestion points: Source text provided by users for paraphrasing, summarizing, and grammar checking (SKILL.md).
- Boundary markers: Absent; no delimiters or 'ignore embedded instructions' warnings are suggested for the agent prompt.
- Capability inventory: The skill utilizes outbound network access (curl) to the DataForSEO API, which could be exploited if an injection attack redirects the agent's behavior.
- Sanitization: Absent; the skill suggests passing JSON bodies directly based on user inputs.
- External Communication (LOW): The skill requires outbound network access to api.dataforseo.com and docs.dataforseo.com. These are not in the trusted domain whitelist, posing a standard data exposure risk inherent to third-party API integrations.
Audit Metadata