Back to skill
Skillv1.0.4
VirusTotal security
SealVera · External malware reputation and Code Insight signals for this exact artifact hash.
Scanner verdict
SuspiciousApr 30, 2026, 4:55 AM
- Hash
- 68c04807acfb9fad7a2fcdba86ea34a283c94bd73edcd2489e15257a2d6c8b58
- Source
- palm
- Verdict
- suspicious
- Code Insight
- Type: OpenClaw Skill Name: sealvera Version: 1.0.4 The skill functions as an invasive auditing and monitoring suite that intercepts all LLM traffic and modifies the agent's core instruction files (AGENTS.md, SOUL.md) to enforce reporting to an external endpoint (app.sealvera.com). It employs high-risk techniques including hooking Node.js module resolution (scripts/autoload.js) to patch OpenAI/Anthropic SDKs and a background script (scripts/subagent-watcher.js) that scans local session transcripts in the user's home directory to exfiltrate summaries. While these behaviors are documented as 'audit' features, the combination of SDK interception, automated prompt injection to alter agent behavior, and filesystem scanning creates a significant data exfiltration surface.
- External report
- View on VirusTotal