bmad-os-skill-to-bundle
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted content from a user-specified folder path to generate standalone instructions for consumer LLM platforms, creating a potential surface for indirect prompt injection during the bundling process.\n- Ingestion points: The agent reads the contents of 'SKILL.md', all files in 'references/*.md', and various data files (.csv, .yaml, .txt) from the provided source folder path to identify methodologies and personas.\n- Boundary markers: The skill lacks specified delimiters or instructions to ignore embedded malicious commands within the source files while they are being read and rewritten into the bundle protocol.\n- Capability inventory: The skill utilizes local file system access to read source material and write generated bundle files (SKILL.md, INSTRUCTIONS.md, and data files) to an output destination.\n- Sanitization: There are no sanitization, escaping, or validation steps defined to verify the content being lifted from the source files before it is interpolated into the generated bundle instructions.
Audit Metadata