The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to write, refactor, and test Elixir code, creating a surface where malicious instructions could be embedded in the processed files.
Ingestion points: The agent reads and modifies Elixir source files (.ex) and test files (.exs).
Boundary markers: No explicit markers are defined to separate untrusted file content from instructions.
Capability inventory: The skill utilizes mix compile, mix test, mix format, and mix credo to execute or analyze code.
Sanitization: The instructions do not specify sanitization or validation routines for external code before processing.
[COMMAND_EXECUTION]: The skill guides the agent to use shell commands for project maintenance and verification.
It instructs the use of unbuffer to run mix commands (e.g., unbuffer mix test, unbuffer mix compile) to maintain ANSI output.
It identifies the expect package (containing unbuffer) as a prerequisite and suggests installation via brew install expect or apt install expect.

elixir