The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses mkdir -p and mv to create archive directories and move completed change files. These operations are scoped to the local project's beat/ directory.
[PROMPT_INJECTION]: Evaluation of Indirect Prompt Injection surfaces (Category 8):
Ingestion points: The skill reads several files from the beat/changes/ directory, including status.yaml, tasks.md, proposal.md, design.md, and Gherkin feature files.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to ignore potential instructions embedded in these files.
Capability inventory: File system access (mkdir, mv), local file reading, updating scenario paths in test files, and invoking subsequent skills.
Sanitization: No explicit sanitization or validation of the processed file content is performed. Given the developer-centric nature of the tool and its localized scope, this is considered a standard operational risk.
[SAFE]: The skill's instructions focus on maintaining documentation consistency (BDD sync) and ensuring proper workflow completion, which aligns with its stated purpose.

archive