The Agent Skills Directory

[SAFE]: The skill implements a structured protocol for time-constrained task management, emphasizing validation and hard deadlines. The overall design prioritizes user-defined constraints and verifiable outcomes.
[COMMAND_EXECUTION]: The skill includes a mise.toml configuration that runs local Python scripts (check_code_rules.py, check_evals.py, lint_writing.py, validate_skill.py) for static analysis, linting, and quality assurance. These scripts utilize Python standard library modules (e.g., ast, pathlib, json) to scan skill files and do not perform network operations or access sensitive system paths.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied task descriptions via the /timebox command, which represents a potential ingestion point for untrusted instructions. However, the skill incorporates significant mitigation measures:
Ingestion points: User-supplied task and duration parameters via the /timebox command (SKILL.md).
Boundary markers: Explicit instructions to define the complete outcome, scope, unknowns, and forbidden outcomes before admission (SKILL.md).
Capability inventory: The skill wraps the agent's existing toolset and requires always-current-datetime for clock anchoring.
Sanitization: Employs a 'logic-audit' step to check for conflicting scope or impossible deadlines prior to execution (SKILL.md).

timebox