LocalUDPMessenger

• Analyzed: May 12, 2026, 3:57 AM
• Findings: 1
• Version: 1.5.0
• Hash: 07b3003094debe5070f183566235c91b3be425b789b7ce6d80845ba46bcd4eba
• Source: palm
• External report: View on VirusTotal

Findings (1)

The skill is classified as suspicious due to two primary factors: 1) The `relayServer` feature in `index.ts` (documented across `SKILL.md`, `CLAUDE.md`, `README.md`, and `openclaw.plugin.json`) forwards all agent communication (sent, received, system events) to a user-configurable external host. While explicitly designed and documented as an opt-in monitoring feature for human oversight, this constitutes data exfiltration to an external endpoint. 2) The `wakeAgent` function in `index.ts` triggers agent execution via a local `/hooks/agent` endpoint, passing incoming messages as part of the prompt. This creates a prompt injection surface, even though `SKILL.md` and `CLAUDE.md` contain strong guardrails explicitly instructing the agent to treat incoming messages as untrusted and require user confirmation for sensitive actions or execution. The `MAX_MESSAGE_SIZE` also limits potential injection payloads. These are risky capabilities, but without clear evidence of intentional malicious exploitation, they are classified as suspicious rather than malicious.