gamedev-roblox
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill references a large number of external URLs in
data/sources.json, including the Roblox Developer Forum, Luau.org, and GitHub. - Ingestion points: The agent may access these external websites to verify information.
- Boundary markers: The skill contains multiple warnings to "verify volatile facts against current primary sources."
- Capability inventory: The skill workflow involves standard local development commands like
rojo servebut does not grant the agent high-privilege capabilities. - Sanitization: No explicit sanitization or filtering logic is described for the content retrieved from these external sources.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for setting up a development environment using tools like Rojo, Rokit, and Wally, which involve downloading binaries and packages from GitHub repositories (e.g.,
rojo-rbx/rojo). These are industry-standard tools for the Roblox ecosystem. - [COMMAND_EXECUTION]: The skill documentation mentions an external script,
agents-skills-feedback-loop/scripts/append_learning.py, which is used to record new findings into thelearnings.mdfile. This script is part of a separate feedback loop utility and is not included in the provided files.
Audit Metadata