writing-beats

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a creative writing tool that manages a structured 'beat' system for article construction. It interacts with the local file system only as directed by the user to save and update article progress.
  • [DATA_EXFILTRATION]: No network-capable tools or patterns were found. The skill does not attempt to send data to external servers or domains.
  • [PROMPT_INJECTION]: The instructions are clear and focused on the writing process. There are no attempts to bypass safety filters, extract system prompts, or override agent behavior.
  • [COMMAND_EXECUTION]: The skill does not utilize shell commands, subprocesses, or dynamic code execution. It relies on standard file I/O to manage markdown content.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided markdown files ('raw material'). While this represents a surface for indirect instructions, the agent's workflow is strictly constrained to paraphrasing and organizing content into beats, and the agent lacks the high-privilege capabilities (network, shell) required for a significant compromise. (Severity: LOW).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 08:21 PM
Security Audit — agent-trust-hub — writing-beats