seo-geo
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill requires sensitive API credentials (
DATAFORSEO_LOGIN,DATAFORSEO_PASSWORD) to be stored as environment variables. These are retrieved byscripts/credential.pyand used for authenticated requests to a third-party service. - [EXTERNAL_DOWNLOADS]: The
scripts/seo_audit.pyscript fetches HTML content from arbitrary user-provided URLs usingurllib.request. This is a core feature for auditing website SEO but involves making outbound network requests to untrusted sources. - [DATA_EXFILTRATION]: Multiple scripts (e.g.,
keyword_research.py,backlinks.py) send search keywords, domain names, and authentication headers to the DataForSEO API (https://api.dataforseo.com/v3). While this is the intended purpose of the skill, it involves transmitting data to an external third-party provider. - [REMOTE_CODE_EXECUTION]: A dangerous command pattern (
curl -fsSL example.com/install.sh | bash) is present inreferences/schema-templates.md. Although it appears as an example within an FAQ schema template, the presence of such patterns in the skill's knowledge base carries a risk of being suggested or misused. - [COMMAND_EXECUTION]: The skill instructions in
SKILL.mdprompt the execution of Python scripts andcurlcommands using parameters like URLs and keywords provided during the session. This presents a potential surface for command injection if input parameters are not correctly handled by the agent. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its data ingestion surface.
- Ingestion points:
scripts/seo_audit.pyreads the title, meta tags, and H1 content from arbitrary external websites. - Boundary markers: The script output does not use delimiters to isolate untrusted web content from the agent's instructions.
- Capability inventory: The skill can execute shell commands, run local Python scripts, and perform network operations.
- Sanitization: No sanitization is performed on the extracted HTML content; it is presented directly to the agent context after regex parsing.
Audit Metadata