The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a --verification-command argument within the SSF workflow complete command. This parameter is designed to execute arbitrary shell commands provided by the agent or user to verify the completion of a task.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection as it actively processes untrusted external data to drive its workflow.
Ingestion points: The workflow relies on reading and interpreting content from proposal.md (for scope and risks) and tasks.md (for ordered tasks and proof) within the local repository.
Boundary markers: There are no instructions to use boundary markers or to ignore malicious instructions that might be embedded within these markdown files.
Capability inventory: The skill possesses the capability to execute arbitrary shell commands (via the verification flag) and perform file system operations (creating directories under changes/).
Sanitization: The instructions do not mandate any sanitization or validation of the data extracted from the project files before it is used to populate command-line arguments or influence agent behavior.

workflow-start