The Agent Skills Directory

[SAFE]: The skill provides a highly structured framework for iterative planning and architecture work. The instructions focus on improving plan quality through adversarial lenses (feasibility, operability, risk) and explicit convergence gates.
[INDIRECT_PROMPT_INJECTION]: The skill includes a specific defensive instruction to treat external inputs as untrusted context, reducing the risk of indirect prompt injection attacks from imported documents.
[COMMAND_EXECUTION]: The provided Python script plan_contract_lint.py is a standalone utility that uses standard libraries to perform regex-based validation of markdown files. It does not perform network operations or access sensitive system files.
[METADATA_POISONING]: The YAML frontmatter and skill description accurately reflect the skill's purpose as a decision-complete planning tool.

plan