mc-develop-first
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the download and installation of Minecraft development resources (MDM packages) using the
mdmReleaseInstalloperation. The instructions require the agent to inspect recommendations first and only proceed with the download if the user has provided explicit authorization for thedownloadPolicy: "allowed"setting. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from multiple sources, including local crash logs, user-written Minecraft scripts, and remote mod metadata from Modrinth or CurseForge. While this creates a vulnerability surface for indirect prompt injection, the instructions mitigate this by emphasizing structured tool fields and evidence-based triage.
- Ingestion points: Minecraft crash logs, KubeJS/ProbeJS scripts, and remote metadata from mod repositories.
- Boundary markers: The skill utilizes structured JSON inputs for all tool interactions, which helps isolate the agent's control logic from the data being processed.
- Capability inventory: The associated MCP tool (
mc_develop) provides capabilities for reading files, inspecting jar archives, and installing resource packages. The agent is also instructed to use standard filesystem tools for editing. - Sanitization: The skill relies on the structured output of the
mc_developtool rather than raw text interpolation of external content into the primary prompt instructions.
Audit Metadata