The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill processes external specifications to generate implementation plans, which creates a potential vector for instructions in the input data to influence the generated code or commands.
Ingestion points: Processes user-provided "spec or requirements" for tasks.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore embedded instructions within the source specs.
Capability inventory: The resulting plans include shell commands (pytest, git commit) and Python code blocks that are intended for execution by the agent or subsequent sub-skills.
Sanitization: The skill does not implement sanitization or validation logic for the input specs.
[Command Execution] (LOW): The skill explicitly templates the generation of shell commands (pytest and git). While these are standard development tools, they represent the capability to execute arbitrary logic if the generated plan is modified by malicious input.

writing-plans