The Agent Skills Directory

[SAFE]: The skill facilitates the use of CipherStash encryption tools and libraries. All external dependencies like @cipherstash/stack and @cipherstash/stack-drizzle are legitimate resources provided by the skill author.
[SAFE]: Sensitive configuration is managed through environment variables (DATABASE_URL), adhering to standard security protocols for database access.
[COMMAND_EXECUTION]: The skill provides various CLI commands for database setup, migrations, and encryption management (e.g., stash eql install, stash encrypt backfill). These are standard operational commands for the CipherStash toolset and do not involve privilege escalation or high-risk execution patterns.
[EXTERNAL_DOWNLOADS]: The skill utilizes the vendor's command-line interface and package managers to install and update necessary encryption tools. These downloads are from the vendor's own infrastructure and registries.
[INDIRECT_PROMPT_INJECTION]: The skill defines an interface for reading external data from a PostgreSQL database, which constitutes a potential attack surface for indirect prompt injection. However, the skill provides robust boundary markers and validation.
Ingestion points: Database select queries demonstrated in SKILL.md.
Boundary markers: Usage of extractEncryptionSchema to enforce type safety and data structure.
Capability inventory: Database CRUD operations and field-level encryption/decryption client methods.
Sanitization: Encryption operators implement specific validation logic, such as rejecting short needles in search queries or empty objects in JSON containment checks.

stash-drizzle