grep-build-app
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple local Node.js scripts (
update-check.js,auth.js,grep-api.js) for authentication, background update checks, and job management using the shell. - [DATA_EXFILTRATION]: The skill gathers local project information, including file contents from
package.json,pyproject.toml, and user-specified data files (CSV/JSON), and transmits this data to the external endpoint atapi.grep.aifor processing by an app-builder service. - [EXTERNAL_DOWNLOADS]: Upon completion of a build job, the skill downloads generated HTML, CSS, and Javascript assets from the remote service to a local directory (
/tmp/grep-app-<slug>). - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes external, untrusted content from the project environment to build the 'product brief' sent to the AI expert.
- Ingestion points: Processes local project files, configuration files (
package.json,pyproject.toml), and user-provided URLs in Step 3. - Boundary markers: No specific delimiters or 'ignore' instructions are used to wrap the gathered context files.
- Capability inventory: Executes shell commands via Node.js scripts, performs network requests to the Grep API, and writes files to the local temporary directory.
- Sanitization: No evidence of sanitization, validation, or escaping of the gathered project context before interpolation into the build prompt.
Audit Metadata