The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill generates Elixir source code at runtime using templates and user-provided OpenAPI specifications. It then proceeds to execute this generated code using mix compile and mix test. If the input specification is maliciously crafted, it could result in arbitrary code execution within the developer's environment during the build or test phase.
[DYNAMIC_EXECUTION]: The Schemas template in references/templates.md includes a from_map/1 function that uses String.to_atom(key) on keys from JSON input at runtime. In Elixir, atoms are not garbage collected, and creating them from untrusted input is a known security vulnerability (Atom Exhaustion) that can lead to a Denial of Service (DoS).
[COMMAND_EXECUTION]: The skill executes multiple shell commands to manage the Elixir project lifecycle, including mkdir, mix deps.get, mix compile, mix test, and git init. The mix commands involve running the generated project's configuration and dependencies.
[EXTERNAL_DOWNLOADS]: The skill is designed to fetch OpenAPI specifications from external URLs provided by the user. Additionally, the mix deps.get command downloads external Elixir packages from the Hex.pm registry.
[INDIRECT_PROMPT_INJECTION]: The skill lacks sanitization for fields extracted from the OpenAPI specification (such as schema names, descriptions, or operation summaries) before they are interpolated into Elixir quote blocks or used in Module.concat calls. This creates a surface for indirect prompt injection where a malicious spec could inject code to escape the intended logic.

bootstrap-elixir-openapi