paper-spine-build
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates several local Python utility scripts to perform its duties. It demonstrates a security-conscious approach by explicitly requiring the agent to quote directory paths and avoid shell globbing, protecting the system from command injection vulnerabilities when working with user-controlled directory structures.
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and synthesize data from an untrusted materials directory (ingestion points). Although the skill does not use explicit boundary delimiters (boundary markers), it controls the agent's behavior through a rigorous writing rationale matrix and evidence bank that mandate factual grounding (sanitization). The skill's capabilities are limited to local file system writes and the execution of included scripts, which prevents broader exploitation (capability inventory).
Audit Metadata