chapter-lead-writer
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill provides instructions for maintaining writing style, thematic coherence, and structural constraints. It does not contain any patterns typical of prompt injection, such as attempts to bypass safety filters or ignore prior instructions.
- [DATA_EXFILTRATION]: The skill reads from local project files (YAML, JSONL, BIB) and writes to local markdown files within a designated directory. It specifies 'Network: none' and contains no commands that would facilitate data exfiltration to external domains.
- [REMOTE_CODE_EXECUTION]: The skill is purely instructional for text generation and does not include any scripts, package installations, or remote code patterns.
- [COMMAND_EXECUTION]: No shell commands, system calls, or subprocess executions are defined within the skill workflow.
- [EXTERNAL_DOWNLOADS]: The skill does not request or perform any downloads from external sources.
- [CREDENTIALS_UNSAFE]: No hardcoded API keys, tokens, or credentials were found in the skill metadata or body.
Audit Metadata