The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the AI to execute local Python scripts using the uv run command, specifically targeting scripts like resolve_customization.py and resolve_config.py. These scripts are expected to reside within the project's {project-root}/_bmad/scripts/ directory.
[DYNAMIC_EXECUTION]: The workflow includes the execution of arbitrary shell commands defined in configuration files (customize.toml) under the activation_steps and on_complete keys. This mechanism allows for environmental extension but executes commands based on potentially untrusted project configuration.
[INDIRECT_PROMPT_INJECTION]: The skill ingests content from external sources, such as existing project documentation and web research results, and uses this information to generate the final PRFAQ document.
Ingestion points: The skill scans documents in {planning_artifacts}, {project_knowledge}, and user-provided paths using the Artifact Analyzer subagent.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the ingested text before processing it.
Capability inventory: The skill possesses capabilities to execute shell commands, write to the file system, and perform network operations through subagents.
Sanitization: While the Artifact Analyzer returns data in a structured JSON format, providing a layer of schema validation, no further sanitization or content filtering is performed on the raw text insights extracted from the documents.

bmad-prfaq