The Agent Skills Directory

[PROMPT_INJECTION]: The skill's metadata incorrectly identifies the author as 'vercel', which contradicts the actual author identity of 'the-utopia-studio'. This deceptive labeling can lead to a misjudgment of the skill's provenance and trustworthiness.
[DATA_EXFILTRATION]: The skill facilitates the transfer of project files and configuration data to external endpoints (vercel.com and vercel.sh). While these are official domains for the deployment service, the broad packaging of project contents constitutes a data exposure risk.
[PROMPT_INJECTION]: The skill ingests untrusted local project data to automate the deployment process, creating a surface for indirect prompt injection attacks.
Ingestion points: Local project files, specifically package.json, are read to detect the project framework within the deployment scripts.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to prevent the agent from executing commands found within the project data.
Capability inventory: The skill uses shell commands including git, cat, tar, curl, npm, and vercel to interact with the filesystem and network.
Sanitization: The deployment scripts include specific exclusions for .env and .git directories to protect credentials, but no generic sanitization is applied to other project files.
[COMMAND_EXECUTION]: The skill makes extensive use of system commands and sub-processes to manage project state, package the directory, and perform network uploads.

deploy-to-vercel