The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it incorporates untrusted data from user configuration and local file system discovery.
Ingestion points: The discovery phase (Phase 0) audits the existing application structure, and the configuration phase (Phase 1) accepts user-defined credit pack keys and cost mappings.
Boundary markers: The skill instructions do not specify the use of clear delimiters or instructions to the agent to treat data gathered from the file system or user input as untrusted content during code generation.
Capability inventory: The skill has broad capabilities including generating and writing Elixir source files, modifying project dependencies in mix.exs, and executing shell commands for database migrations.
Sanitization: There is no explicit requirement for the agent to sanitize or escape user-provided configuration values before they are interpolated into the generated Elixir code.
[EXTERNAL_DOWNLOADS]: The skill incorporates external libraries and third-party resources into the project environment.
Fetches the pgflow background processing library from the vendor's GitHub repository.
Adds a dependency on the stripity_stripe community package for Elixir.
The JavaScript hook dynamically loads the official Stripe.js library from Stripe's Content Delivery Network (https://js.stripe.com/v3/) at runtime.
[COMMAND_EXECUTION]: The skill requires the execution of shell commands to manage project state and verify the implementation.
Instructs the agent to run mix deps.get to retrieve external project dependencies.
Uses mix pgflow.gen.job_migration to generate the necessary database schemas for credit fulfillment jobs.
Executes mix ecto.migrate to apply schema changes to the application's database.
Runs mix test to validate the integrated functionality.

bootstrap-credits