deploy-to-vercel
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local shell commands and scripts to manage the deployment workflow. It uses git for repository interactions and deployment triggers, and invokes the vercel CLI for linking, inspection, and direct deployment. Internal helper scripts (resources/deploy.sh and resources/deploy-codex.sh) package and upload projects using tar and curl. \n- [EXTERNAL_DOWNLOADS]: Installs the Vercel CLI from the npm registry using npm install -g vercel to ensure required tools are available. \n- [DATA_EXFILTRATION]: Transfers project source code to Vercel's deployment infrastructure at claude-skills-deploy.vercel.com and codex-deploy-skills.vercel.sh. The skill implements security controls by explicitly excluding .env, .git, and node_modules from uploaded archives. \n- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes untrusted local files. \n
- Ingestion points: Reads package.json for framework detection and traverses the project directory for packaging. \n
- Boundary markers: None present. \n
- Capability inventory: File read, network upload (curl), command execution (git, vercel). \n
- Sanitization: Implements file path exclusion for known sensitive environment variables and version control metadata.
Audit Metadata