se-dev-mod
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The preparation scripts (
Prepare.batandcommon-posix.sh) download and execute installation scripts for theuvtool fromastral.sh.- [EXTERNAL_DOWNLOADS]: The skill fetches thebusyboxexecutable fromfrippery.orgto provide a UNIX-like environment on Windows systems.- [COMMAND_EXECUTION]: The skill requests broad permissions to execute shell commands and Python scripts as part of its indexing and search workflow.- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by indexing source code from untrusted third-party mods (e.g., from the Steam Workshop). Malicious instructions embedded in these files could potentially be ingested and acted upon by the agent when searching or reading code. - Ingestion points: Mod source files (
.cs) parsed by the indexing logic. - Boundary markers: No delimiters or safety instructions are used when presenting searched content to the agent.
- Capability inventory: The skill allows execution of shell commands and interaction with a local code index.
- Sanitization: No sanitization or filtering of the indexed mod content is performed.
Audit Metadata