exam-forecast
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill processes untrusted external data in the form of past exams provided as PDFs or text. This represents a potential surface where embedded instructions in the source documents could attempt to influence the agent's analysis.
- Ingestion points: Past exams are ingested in Step 3 of the workflow via PDF, pasted text, or file paths.
- Boundary markers: No explicit boundary markers or 'ignore' instructions for the intake phase are defined in the skill.
- Capability inventory: The skill possesses file-reading and file-writing capabilities within the
~/.claude/plugins/config/claude-for-legal/directory. - Sanitization: The skill does not specify explicit sanitization or filtering of the content extracted from the exams, relying instead on a structured analysis workflow.
- Local File System Operations: The skill reads from and writes to specific paths in the
~/.claude/directory to maintain state and save forecast outputs. This localized file access is a standard mechanism for persistence and context management within the agent's workspace.
Audit Metadata