The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes arbitrary shell commands defined in task tracking files (e.g., 'TODO.csv', 'SUBTASKS.csv', and 'SPEC.md') under fields such as 'validation_command' and 'Final Validation Command' to verify milestone completion.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection within its Batch Task workflow (as defined in 'assets/BATCH_TEMPLATE.md'). This workflow interpolates row-level data from 'workers-input.csv' into an instruction template for sub-agents. Ingestion points: Data enters the context via 'workers-input.csv', 'TODO.csv', and 'SUBTASKS.csv'. Boundary markers: There are no explicit delimiters or 'ignore' instructions used to encapsulate interpolated variables in the templates. Capability inventory: The skill allows spawning sub-agents via 'spawn_agents_on_csv' and executing shell commands via 'validation_command' (referenced in 'SKILL.md' and 'assets/todo_template.csv'). Sanitization: The protocol does not specify any validation, escaping, or sanitization of the CSV-sourced data before it is processed by the agent.

taskmaster