JARVIS AI Skills
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: jarvis-skills Version: 1.0.0 The skill bundle appears benign based on the provided metadata and documentation. The `SKILL.md` and `skill.json` files clearly define the purpose as robotic arm and gripper manipulation, listing appropriate dependencies (`openclaw`, `pyserial`, `numpy`) and interfaces (`grab`, `release`, `move_to`). There is no evidence of prompt injection attempts against the agent, data exfiltration, malicious execution, or other harmful behaviors in the analyzed files. The instructions and descriptions are consistent with the stated purpose of controlling a physical robot.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If connected to real hardware, mistaken or unintended commands could move a robot arm, grab or release objects, damage property, or injure someone nearby.
The skill describes direct physical robot and gripper operations, including sequenced actions, but does not define confirmation prompts, safety interlocks, workspace limits, or force/speed constraints for these high-impact actions.
- Robotic arm movement (6-DOF) - Gripper grab/release operations - Action sequence execution ... - "Jarvis, grab the object" - "Jarvis, move to 10 20 30"
Use only in simulation or a controlled test environment until the skill adds explicit user approval, device allowlisting, emergency-stop guidance, workspace boundaries, and conservative force/speed defaults.
Users may need to obtain or create the missing control file separately, meaning the actual code that talks to robotic hardware may not be the code that was reviewed here.
The package declares a required control module, but the provided manifest contains only skill.json and SKILL.md and there is no install spec, so the core implementation that would control hardware is absent from review.
"requiredFiles": ["openclaw_control.py"]
Require the complete implementation, pinned dependencies, and installation instructions to be included and reviewed before using this skill with any real robot.
A user could over-trust the stated safety features and use the skill around real hardware before verifying that collision detection, limits, and emergency controls actually work.
The skill makes broad safety and performance claims for powerful robotic hardware, but the reviewed package does not include the implementation needed to substantiate those claims.
- Collision detection and safety - Hardware auto-detection ... - Payload: 3-500 kg (model-dependent) - Speed: 1-7000 mm/s
Treat the safety and performance claims as unverified until tested against the specific robot model with independent safeguards in place.