dataforseo-business-data-api
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [Metadata Poisoning] (MEDIUM): The skill metadata claims it was generated with 'OpenAI GPT-5.2'. This version of the model does not exist, indicating misleading or hallucinated metadata that misrepresents the skill's origin and potentially its reliability.
- [Indirect Prompt Injection] (MEDIUM): High risk of adversarial content ingestion.
- Ingestion points: Data enters the agent context from
api.dataforseo.com, specifically through business listings, Google reviews, and Questions & Answers. - Boundary markers: The instructions do not define boundary markers (e.g., XML tags or delimiters) to separate the untrusted external data from the agent's instructions.
- Capability inventory: The skill has the capability to perform outbound HTTP requests and process raw JSON payloads to generate summaries or alerts for the user.
- Sanitization: No sanitization or filtering logic is provided to handle potentially malicious instructions embedded in reviews or Q&A fields.
- [Prompt Injection] (LOW): The 'Examples' section contains instructional prompts such as 'If you don't have the skill installed, install dataforseo-business-data-api and then continue', which are designed to coerce agent behavior into installing external code.
- [Credentials Unsafe] (LOW): The skill relies on HTTP Basic Authentication using
${DATAFORSEO_LOGIN}and${DATAFORSEO_PASSWORD}. While necessary for the API, it requires the agent to handle sensitive credentials in its environment.
Audit Metadata