The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands defined as 'steps' within the plan.md file. It also performs git operations such as git log, git add, and git commit to manage project state.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes external data from the project environment, which could potentially be manipulated to influence the agent's behavior.
Ingestion points: The agent reads control files including .beyond-code/plan.md, .beyond-code/gate.md, and .beyond-code/config.yaml. Additionally, the 'First-Principles Root-Cause Protocol' requires the agent to perform deep reads of source code and data contracts.
Boundary markers: The skill does not explicitly use delimiters or specialized instructions to ignore potential natural language commands embedded within the code or markdown files it processes.
Capability inventory: The skill has significant capabilities including arbitrary shell command execution (via task steps), local file modification (gate.md), and repository management (git).
Sanitization: There is no evidence of content sanitization or validation of the task 'steps' before they are passed to the shell for execution.

beyond-code-build