writing-beats
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests raw material from user-provided markdown files and writes content to a specified article file. This establishes a surface where malicious instructions in the input material could attempt to influence the agent. However, the risk is mitigated by the skill's design, which requires the agent to present candidates and obtain user approval for every beat before writing to disk.
- Ingestion points: Markdown files provided by the user (SKILL.md).
- Capability inventory: Reading from and writing to the local filesystem for the article draft (SKILL.md).
- Boundary markers: No specific delimiters or "ignore instructions" markers are defined for the raw material content.
- Sanitization: No automated validation or filtering of the input material is specified in the static instructions.
- [SAFE]: The skill's frontmatter configuration explicitly disables model invocation, which is a security best practice that limits the agent's ability to perform autonomous actions or access other tools during the writing process.
Audit Metadata