strategic-review-interactive

Pass

Audited by Gen Agent Trust Hub on Feb 27, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes user-provided markdown files without protective delimiters. \n
  • Ingestion points: Strategic reports are retrieved based on user input in SKILL.md and their content is processed in interactions/init.md, interactions/approve.md, and interactions/revision.md. \n
  • Boundary markers: None identified; the agent is not instructed to disregard instructions found within the processed reports. \n
  • Capability inventory: Includes the ability to call the strategic-review skill and save/update files on the system. \n
  • Sanitization: No sanitization or validation of the report content is performed before use. \n- [DATA_EXFILTRATION]: The skill allows the user to specify filenames for retrieval in SKILL.md (Step 1). Without directory-level restrictions or strict path validation, this creates a vulnerability surface for path traversal, potentially allowing access to sensitive files that match the markdown format requirement.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 27, 2026, 08:24 AM