The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements authentication flows that ingest untrusted user data. Ingestion points: Data enters the agent context through email and password fields in SKILL.md and references/password-flows.md. Boundary markers: Input is delimited by standard JavaScript function parameters and object properties. Capability inventory: The skill performs database writes via createUserAsync and upsertAsync, and sends emails via sendVerificationEmail and related methods in SKILL.md and references/password-flows.md. Sanitization: External content is sanitized by Meteor's built-in accounts and password packages, which utilize bcrypt hashing and secure token validation.
[EXTERNAL_DOWNLOADS]: The skill references official documentation and package source code hosted on GitHub under the meteor organization. These references are used for developer guidance and do not involve dynamic execution of untrusted scripts.
[COMMAND_EXECUTION]: A developer utility command using meteor node is provided in references/oauth-services.md to generate a random 16-byte key for OAuth encryption. This is a local administrative task for configuration and does not present a runtime security risk.

meteor-accounts