compute-env-setup
Pass
Audited by Gen Agent Trust Hub on Aug 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of shell commands to automate environment setup, including package management (conda, pip), container operations (apptainer), and repository management (git). It also employs
sedfor programmatic patching of legacy source code to resolve deprecated library aliases (e.g., updating NumPy 1.24 compatibility). - [EXTERNAL_DOWNLOADS]: The skill fetches numerous software packages, model weights, and datasets from well-known scientific and research infrastructure providers, including GitHub, HuggingFace, the RCSB Protein Data Bank, and specialized research repositories like PyG (data.pyg.org).
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its reliance on the
compute_detailstool, which stores environment configurations in a persistent markdown ledger. If this shared ledger were compromised, an agent could ingest misleading instructions during environment setup. However, the skill includes robust validation protocols, such as kernel-dispatch witnesses and agent-driven documentation verification, to mitigate these risks.
Audit Metadata