writing-beats
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process external markdown files provided by the user to extract content for writing "beats."
- Ingestion Points: The agent is instructed to read "raw material" from a markdown file provided by the user (SKILL.md, under "what-to-do").
- Capability Inventory: The skill possesses file read and write capabilities, as it must read the raw material and the evolving article from the disk and write new "beats" to the article file (SKILL.md, Steps 3, 4, and "Writing Rhythm").
- Boundary Markers: The instructions lack explicit boundary markers or directions to the agent to treat the content of the raw material strictly as data, leaving it vulnerable to instructions embedded within the source markdown (e.g., instructions intended to hijack the agent's writing style or behavior).
- Sanitization: There are no defined procedures for sanitizing, escaping, or validating the external content before the agent paraphrases or quotes it.
Audit Metadata