review-lesson
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes content from both local files and external web pages without using boundary markers or sanitization. Malicious instructions embedded in a lesson file or a compromised documentation page could potentially influence the agent's behavior.
- Ingestion points: Local lesson files (
learn-claude/*.md) and official Anthropic documentation (docs.anthropic.com). - Boundary markers: Absent. No delimiters or instructions are used to separate the ingested content from the agent's internal logic.
- Capability inventory: File system read access and network GET requests defined in SKILL.md.
- Sanitization: Absent. The skill does not provide methods for escaping or validating the content retrieved from external or local sources.
- [DATA_EXFILTRATION]: The skill allows reading local files based on user-provided arguments. While designed to read lesson files, this capability could be used to attempt access to other sensitive local files if the agent's environment does not enforce strict path limitations or sandboxing.
- [EXTERNAL_DOWNLOADS]: The skill fetches documentation from Anthropic's official domain for cross-referencing. This involves network requests to a well-known service for technical reference purposes.
Audit Metadata