motioneyes-visual-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted visual data such as frame sequences and video files to generate summaries and metrics. An attacker could potentially embed malicious instructions within these visual assets to influence the agent's behavior during the interpretation phase.
- Ingestion points: SKILL.md defines inputs from frame directories, video files, and macOS window captures.
- Boundary markers: The instructions do not define specific delimiters or warnings to ignore embedded content in visual data.
- Capability inventory: The skill executes local scripts (scripts/analyze_sequence.py, scripts/capture_sim_frames.py) and system tools (screencapture).
- Sanitization: No explicit sanitization or validation of the visual content is mentioned before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill requires the execution of several shell commands and local Python scripts to capture frames and perform computer vision analysis.
- Evidence: SKILL.md and references/grid-overlay-notes.md instruct the agent to run python3 scripts/analyze_sequence.py, python3 scripts/capture_sim_frames.py, screencapture, and git submodule update.
- [EXTERNAL_DOWNLOADS]: The skill uses a git submodule (third_party/GridGPT) which involves downloading code from an external repository during the setup process.
- Evidence: references/grid-overlay-notes.md contains instructions to run git submodule update --init --recursive.
Audit Metadata