runpod-vllm-gpu-deploy
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references documentation and source code hosted on the author's GitHub repository (
github.com/1solomonwakhungu/runpod-vllm-deployer). These references represent standard development links to the vendor's own infrastructure. - [COMMAND_EXECUTION]: The skill utilizes local shell commands to manage Python virtual environments, install the local package (
pip install .), and execute the deployment CLI (runpod-vllm). These operations are necessary for the skill's primary function and are performed within the local repository context. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted user parameters (e.g., model IDs, environment variables, GPU profiles) to generate deployment plans.
- Ingestion points: User-provided deployment parameters in
SKILL.mdexamples and command arguments. - Boundary markers: Implements a "plan-first" workflow requiring an offline dry-run and explicit user confirmation before any billable or network-enabled deployment occurs.
- Capability inventory: Perform network operations (RunPod API calls), write local state files (
.runpod-vllm-state.json), and execute shell commands. - Sanitization: Includes input validation, such as sanitizing pod names to 63 characters and validating numeric ranges for GPU counts and memory allocations.
Audit Metadata