The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell scripts (scripts/lint-check.sh, scripts/pre-commit-check.sh, scripts/check-coverage.sh) to perform automated quality checks. These scripts interact with standard development tools such as ESLint, Pytest, and various language-specific test runners. This behavior is consistent with the stated purpose of a developer-focused skill and utilizes existing local infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from story files (docs/stories/STORY-*.md) and project configuration (bmad/project.yaml).
Ingestion points: Story files and project configuration are read from the repository into the agent's context.
Boundary markers: The instructions do not define explicit delimiters or escaping for the content of these files.
Capability inventory: The skill can modify the repository and execute local scripts (via Category 4/10 mechanisms).
Sanitization: There is no explicit sanitization of input text.
Note: While the surface exists, the structured implementation approach and reliance on predefined standards in REFERENCE.md mitigate the risk of unintended behavior.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill includes a proactive security check in scripts/pre-commit-check.sh that scans for patterns associated with secrets (e.g., API keys, tokens) before code is committed. This is a beneficial security feature designed to prevent accidental data exposure.

bmad-developer