Back to skill
Skillv1.0.2
VirusTotal security
Agent-manager-for-AI-planner · External malware reputation and Code Insight signals for this exact artifact hash.
Scanner verdict
BenignApr 29, 2026, 4:09 AM
- Hash
- 5cd479af0d4a66c154dc0b0a23086c7b9c70b0702117b47b897ca5207e4f603c
- Source
- palm
- Verdict
- benign
- Code Insight
- Type: OpenClaw Skill Name: agentmanager Version: 1.0.2 The OpenClaw AgentSkills skill bundle is classified as benign. The code demonstrates a strong focus on security, particularly in `src/lib/safe-fetch.ts`, `src/security/outbound-policy.ts`, and `src/services/tools.ts`. Outbound network calls are strictly controlled by default, requiring explicit allowlists and blocking IP literals, private addresses, and redirects. Built-in tools like `js_eval` are robustly sandboxed to arithmetic-only expressions, and `file_store` prevents path traversal. The `skill.md` file serves as API documentation and does not contain any prompt injection attempts or instructions for an AI agent to perform unauthorized actions. While the system offers configurable options that could weaken security if misconfigured (e.g., `OUTBOUND_ALLOW_ALL=1`, `ALLOW_INSECURE_HTTP_TOOLS=1`), these are configuration risks rather than evidence of intentional malicious design within the skill bundle itself.
- External report
- View on VirusTotal