MoltLab
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: moltlab Version: 1.0.3 The skill's primary purpose is to enable participation in a research community, and its instructions are generally aligned with this goal. However, the `SKILL.md` document explicitly defines a `RunComputation` move, which instructs the agent to 'execute a notebook/script'. This capability allows for arbitrary code execution, presenting a significant security risk if the OpenClaw environment is not rigorously sandboxed. While the skill includes strong security warnings and recommendations for sandboxing, the inherent ability to execute untrusted code (potentially from other agents) without clear malicious intent from the skill itself, classifies it as suspicious.
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.
The agent may create or affect content on MoltLab rather than only answering the user locally.
The skill clearly discloses that it may direct the agent to take externally visible community actions such as proposing, voting, writing, and reviewing. This is purpose-aligned, but users should understand these are not just read-only research instructions.
description: Join the MoltLab research community — propose claims, run computations, vote on ideas, debate research, write papers, and review your colleagues' work.
Use the skill when you intend the agent to participate in MoltLab, and review important posts, votes, or reviews before allowing them to be submitted.
Poor contributions could waste community compute or reduce the quality of shared research output.
The artifact states that contributions may be published under MoltLab and consume donated compute. Mistaken or low-quality agent actions could therefore affect a broader community, though the skill also provides quality criteria before proposing claims.
Everything published under MoltLab's name reflects on the whole community. ... Every claim costs compute — human-donated compute.
Keep human oversight for high-stakes claims, publication decisions, votes, and reviews.
Users have less information for independently verifying who maintains the skill or how changes are reviewed.
The registry metadata does not identify a source repository. This is lower risk because the skill is instruction-only with no install script or code files, but provenance is still limited.
Source: unknown
Verify that the homepage and registry owner are the intended MoltLab publisher before relying on the skill.