deploy-to-vercel
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes standard git commands (git remote, git add, git push) and Vercel CLI commands (vercel deploy, vercel link, vercel inspect) to facilitate project deployments based on the local project state.
- [EXTERNAL_DOWNLOADS]: The skill provides automated instructions to install the official vercel CLI tool from the npm registry using npm install -g vercel when the tool is not found on the user's system.
- [DATA_EXFILTRATION]: Project source code is packaged and uploaded to Vercel's official deployment endpoints at https://claude-skills-deploy.vercel.com/api/deploy and https://codex-deploy-skills.vercel.sh/api/deploy. The skill implements a security best practice by explicitly excluding sensitive files such as .env, .env.*, and .git folders from the deployment tarball to prevent accidental exposure of credentials.
- [INDIRECT_PROMPT_INJECTION]: The skill processes project files to determine deployment configurations.
- Ingestion points: resources/deploy-codex.sh and resources/deploy.sh read package.json and project directories to identify frameworks and prepare uploads.
- Boundary markers: No specific delimiters are used when reading file contents.
- Capability inventory: The skill possesses file read/write capabilities (via tar) and network egress (via curl).
- Sanitization: The packaging logic proactively filters out credential-containing files like .env to mitigate data leakage risks.
Audit Metadata