The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for resolving tenants from untrusted data sources including subdomains, URL parameters, and request headers. This introduces a surface for indirect prompt injection if the agent processes maliciously crafted tenant identifiers.
Ingestion points: request.subdomains.first, params[:account_slug], and request.headers["Authorization"] in SKILL.md.
Boundary markers: Absent. The skill does not explicitly instruct the agent to use delimiters or ignore embedded instructions within these data points.
Capability inventory: The patterns demonstrate the ability to perform database lookups (Account.find_by), modify session configuration (request.session_options), and perform redirects (redirect_to) in SKILL.md.
Sanitization: Absent. While ActiveRecord provides standard SQL injection protection, there is no explicit sanitization for potential prompt injection in the provided patterns.
[EXTERNAL_DOWNLOADS]: The skill recommends the use of standard and well-known Ruby gems including acts_as_tenant, pundit, sidekiq, and devise. These are established libraries within the Rails ecosystem and are appropriate for the skill's functionality.

multi-tenancy