seo-team-the-writer
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically search engine results (SERPs) and competitor web page content via
seoclicommands. \n - Ingestion points: Data is fetched in Phase 1 (SERP analysis, AI overview analysis, content parsing) from external URLs and search results. \n
- Boundary markers: The skill provides explicit instructions to treat user-provided keywords and URLs as literal values and to avoid interpreting instruction-like text. \n
- Capability inventory: The agent can execute shell commands (
seocli) and write files to the local workspace (workspace/seo/). \n - Sanitization: Instructions mandate quoting all values in shell commands and passing them as single arguments to prevent shell expansion. \n
- [COMMAND_EXECUTION]: The skill relies on the execution of multiple
seoclicommands to interact with the DataForSEO API. \n - Evidence: Shell blocks in
SKILL.mdandreference/seocli-commands.mddemonstrate the use ofseocli serp,seocli keywords-data, andseocli on-page. \n - Mitigation: The skill emphasizes that all commands must go through a user-approval flow and provides specific guidance on quoting user-supplied arguments to prevent command injection.
Audit Metadata