The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes requirement data from external fact sheets which may contain malicious instructions designed to manipulate the agent's implementation or verification steps.
Ingestion points: The skill reads fact labels and commands from .facts files using the facts list and facts check commands as described in the 'Process' section.
Capability inventory: The skill is authorized to write implementation code to the filesystem and execute arbitrary shell commands via the facts check utility.
Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore' instructions for the data being processed.
Sanitization: No evidence of validation or sanitization of the input facts before processing or execution is provided.
[COMMAND_EXECUTION]: The skill explicitly uses the facts check and facts edit commands to run validation logic defined within the fact specification files. This allows for the execution of arbitrary commands that are stored as data within the project's requirement files.

facts-implement