The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is configured to execute shell commands for testing and linting, specifically bundle exec rspec, bundle exec rubocop, and bin/rails console. These tools are standard for the described Rails environment and are used within their intended context.
[PROMPT_INJECTION]: The skill processes external data from test failure reports and user requirements, creating a surface for indirect prompt injection. Malicious instructions could potentially be embedded in test outputs to manipulate the agent's logic during code implementation.
Ingestion points: Failing test output from bundle exec rspec and feature descriptions provided by the user.
Boundary markers: None identified; external data is not clearly delimited from instructions.
Capability inventory: Local environment access including shell execution and delegation to subagents with the ability to modify project source code.
Sanitization: No sanitization of test output or user input is described.

implementation-agent