The Agent Skills Directory

[SAFE]: The skill is a well-engineered developer tool that manages per-workspace database branches. It uses restricted file permissions (mode 0600) for sensitive environment files and implements atomic file operations to prevent state corruption.
[COMMAND_EXECUTION]: The script executes local development tools such as Git and ORM CLIs (Prisma, Drizzle) using safe execution methods (spawnSync with argument arrays) rather than shell interpolation. It properly sanitizes dynamic inputs like branch names through slugification to mitigate command injection risks.
[DATA_EXFILTRATION]: Network communication is limited to the official Neon REST API (console.neon.tech) for legitimate resource management. The skill does not exhibit any patterns of unauthorized data collection or exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill handles metadata from the workspace (e.g., branch names, project IDs) but mitigates potential injection risks by using strict ownership receipts stored in private Git metadata and performing extensive validation before executing destructive actions like database purges.

neondb-branch