coherent-writing
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied draft text and interpolates it into prompts for various sub-agents. This creates a surface where malicious instructions hidden within a document could attempt to influence sub-agent behavior. \n
- Ingestion points: User-provided drafts are ingested in
SKILL.mdto be processed through five separate editing passes.\n - Boundary markers: Sub-agent templates in
references/subagent-prompts.mduse triple-double quotes (\"\"\") to delimit user content from instructions.\n - Capability inventory: The skill is restricted to text processing via the
Tasktool; it contains no instructions or capabilities for network operations, shell command execution, or sensitive file access.\n - Sanitization: The skill employs non-negotiable prompt constraints in every sub-agent call to restrict the agent from introducing new claims or changing the intent of the author.
Audit Metadata