comfyui-inventory

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes curl commands to interact with a local ComfyUI server at 127.0.0.1:8188 to retrieve system statistics, registered nodes, and model lists.
  • [COMMAND_EXECUTION]: Metadata in SKILL.md indicates a dependency on pwsh (PowerShell Core) and references a local script scan-inventory.ps1 used for manual cache refreshing.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from the local filesystem and API responses. 1. Ingestion points: Filesystem directories under COMFYUI_PATH and JSON responses from http://127.0.0.1:8188/object_info as described in SKILL.md. 2. Boundary markers: Absent; no delimiters or instructions to ignore embedded commands within ingested data are identified. 3. Capability inventory: The skill uses curl for network requests to localhost and performs filesystem directory scans. 4. Sanitization: Absent; the skill does not describe any sanitization or validation logic for the filenames or node names retrieved before caching them in state/inventory.json.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 02:14 PM