Openclaw Office Hours

Security checks across malware telemetry and agentic risk

Overview

This appears to be a purpose-aligned, instruction-only workflow that reads repository context and writes a design document, with no evidence of hidden code, exfiltration, or unsafe persistence.

Install only in repositories you are comfortable letting an agent inspect. Review any generated design document before keeping or committing it, and avoid sharing logs, commit history, or project docs that contain secrets or private information.

SkillSpector

By NVIDIA

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly directs the agent to read repository documents and run `git log`, which can expose sensitive project context such as secrets, internal roadmap details, commit metadata, and unpublished work without first informing the user or obtaining consent. In an agent setting, silent access to broad repo context violates least-privilege expectations and can disclose more data than is necessary for the immediate task.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill instructs the agent to write a design document into the project directory as a required step, but does not warn the user that this will modify repository files or request confirmation before doing so. Unannounced file writes can overwrite work, introduce unwanted artifacts into version control, and violate user expectations about whether the interaction is advisory versus mutating.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal