seo-page
Warn
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: In step 4b, the skill specifies the execution of local Python scripts (
scripts/gsc_query.pyandscripts/gsc_inspect.py) via shell commands that interpolate the user-provided{target_url}and{config.default_property}directly into the command line arguments. This pattern is susceptible to command injection if the input is not strictly validated before the shell call is initiated. - [PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection as identified by the following evidence chain:
- Ingestion points: Untrusted data enters the agent context through external URLs fetched via the
WebFetchandFirecrawltools in step 6. - Boundary markers: The instructions do not define boundary markers or explicit prompts to ignore instructions embedded within the fetched HTML/Markdown content.
- Capability inventory: The skill has access to shell execution (Step 4b) and file system writes (folder and report creation), which could be targeted by malicious instructions in scraped data.
- Sanitization: No sanitization, escaping, or validation of the retrieved HTML content is described before the agent uses it to synthesize the final SEO verdict.
Audit Metadata