carto-develop-app
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill generates application code and environment configurations based on user-supplied natural language prompts.\n
- Ingestion points: User prompts and Builder Map IDs (referenced in
SKILL.md).\n - Boundary markers: Absent. The instructions do not define delimiters or safety guidelines to separate user input from generated command logic.\n
- Capability inventory: Execution of
cartoCLI commands for credential management andnpmfor dependency installation and local execution (referenced inSKILL.mdandreferences/auth-public-token.md).\n - Sanitization: Absent. No evidence of input validation or escaping before interpolation into shell commands.\n- [COMMAND_EXECUTION]: The skill relies on executing system commands like
carto auth status,carto connections list, andnpm install.\n - Context: These operations use the vendor's own tools and standard registries, which is expected for this skill's function.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates installation of packages such as
@carto/api-clientand@deck.gl/carto.\n - Context: All referenced resources belong to the vendor's infrastructure or well-known services like Auth0 and MapLibre.
Audit Metadata