ghostwriter-mcp
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to retrieve and process data from a Ghostwriter instance, which could contain malicious instructions embedded in project or log metadata.
- Ingestion points: Project details and operation logs retrieved via the MCP tools 'list_projects' and log inspection.
- Boundary markers: Absent; no instructions are provided to the agent to distinguish between data and system prompts.
- Capability inventory: While the skill body is markdown, the agent utilizes tools for project listing and credential verification.
- Sanitization: Absent; the skill does not specify validation or sanitization for the retrieved data before processing.
Audit Metadata