wilma
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches the official
@wilm-ai/wilma-clipackage from the npm registry as part of the installation process.\n- [REMOTE_CODE_EXECUTION]: Thescripts/wilma-cli.shscript utilizesnpxto execute the Wilma CLI tool, which may involve downloading and running code from the npm registry at runtime if it is not available locally.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from the Wilma platform that could contain malicious instructions.\n - Ingestion points: Untrusted content enters the agent's context through
wilma news,wilma messages, andwilma homeworkcommands, which retrieve text written by school staff (found in SKILL.md).\n - Boundary markers: No explicit delimiters or instructions are provided to the agent to treat fetched data as untrusted or to ignore instructions embedded within that data.\n
- Capability inventory: The skill allows the agent to execute shell commands via
scripts/wilma-cli.shand perform file system operations such as downloading news attachments.\n - Sanitization: Content retrieved from the Wilma system is processed and presented to the agent without explicit sanitization or validation for adversarial content.
Audit Metadata