Exa Research

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed Exa web-research helper that runs a local Exa CLI wrapper and uses an Exa API key, with no hidden or unrelated behavior found.

Install only if you trust the Exa CLI and Exa service. Keep the API key in OpenClaw secrets or a locked-down file, verify what EXA_BIN or PATH resolves to, prefer scoped commands over raw API calls, and avoid sending secrets, private URLs, or proprietary content unless Exa processing is acceptable.

SkillSpector

By NVIDIA

Vulnerability Patterns

MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
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

Findings (1)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 88% confidence
Finding: The skill instructs the agent to invoke a local shell wrapper (`scripts/exa-with-key.sh`) and describes credential discovery from environment variables and local files, but it declares no corresponding permissions or trust boundaries. This creates a real capability/permission mismatch: an agent may execute shell commands and access sensitive local credentials or arbitrary endpoints via the `raw` mode without users or policy systems being explicitly informed.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal