case-study
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill instructions direct the agent to execute a JavaScript file (
scripts/ari.mjs) to compute readability scores. The logic allows for searching for this file in locations that include user uploads, pasted messages, and project files. This creates a risk where an attacker could provide a malicious script with the same name that the agent executes at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including case studies, teasers, and persona configuration files provided by the user. These inputs enter the agent's context and lack robust boundary markers or sanitization, potentially allowing instructions embedded in the processed text to influence the agent's behavior.
- Ingestion points: Project files, user uploads, and pasted text (case studies, teasers, persona.md, voice.md).
- Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore embedded commands in the user-provided text.
- Capability inventory: The skill possesses the capability to execute shell commands (Node.js) and import local JavaScript modules.
- Sanitization: No sanitization or validation of the ingested content is specified before the agent processes the text or executes related logic.
Audit Metadata