The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute arbitrary shell commands defined under the run and hook keys (before_all, after_all, error) in the .asc/workflow.json configuration file.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it executes instructions (shell commands) provided in a local file that an attacker could potentially modify in a compromised or malicious repository.
Ingestion points: The agent reads and processes automation logic from .asc/workflow.json and CLI parameters.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between trusted instructions and potentially untrusted data within the JSON configuration.
Capability inventory: The skill allows for arbitrary shell command execution, environment variable interpolation, and sub-workflow orchestration.
Sanitization: No specific sanitization or escaping mechanisms are mentioned for parameters interpolated into shell commands via $VAR syntax.

asc-workflow