The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard Elixir/Erlang toolchain commands such as mix deps.get, mix ecto.migrate, and mix test. These are expected operations for a project setup playbook and are performed within the local project context.
[DATA_EXFILTRATION]: The skill includes explicit security controls to prevent data exposure. It mandates that no secrets, tokens, or environment-specific URLs be committed to version control and instructs the user to use .env files and redacted connection strings.
[INDIRECT_PROMPT_INJECTION]: The skill processes project configuration files like .tool-versions and .elixir-version. While these are external inputs that influence agent behavior (version selection), they are standard project metadata and the skill does not interpolate untrusted data into high-privilege execution contexts without human oversight.
Ingestion points: Reads .tool-versions, .elixir-version, and .env.example from the local workspace.
Boundary markers: None explicitly defined in the prompts.
Capability inventory: Executes mix subprocesses for dependency management and database migrations.
Sanitization: None, as it assumes the developer is working on a trusted or controlled codebase.
[SAFE]: The skill implements a 'Human-in-the-Loop' (HITL) requirement for destructive operations such as ecto.drop or force-pushing to a repository, providing a manual safety check against accidental or malicious data loss.

setup