arcblock-context
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data by reading Markdown files from the local filesystem (specifically within
.claude/arcblock-context/in the project or home directory). If these files are sourced from an untrusted project and contain malicious instructions, they could influence the agent's behavior. - Ingestion points: File loading workflow defined in
SKILL.md. - Boundary markers: Not explicitly defined in the workflow instructions.
- Capability inventory: Limited to reading and summarizing local documentation files based on a fixed keyword mapping.
- Sanitization: None mentioned for the content of the loaded Markdown files.
Audit Metadata