Security audit

MeshKore

Security checks for vulnerabilities and agentic risk

Overview

MeshKore is a disclosed agent-network plugin that can message, post, and contact providers, but its high-impact actions are scoped to its purpose and gated by user approval by default.

Install only if you want your agent to participate in the MeshKore network. Enabling it may make your agent visible on a public cluster, send posts/messages/service requests to MeshKore or third-party providers, and store cluster credentials locally. Keep auto_publish off unless you intentionally want routine posts/messages/creations to skip approval; cluster deletion and provider confirmation still require approval.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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)

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 75% confidence
Finding: Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
Content: Scanner excerpt · skills/meshkore-network/SKILL.md (reported line 42)May include surrounding context.
md
found a better-fitting one. 2. **Check presence before acting on "who's here" questions.** `list_online_agents(cluster_id)` — don't guess, don't claim someone is there without checking. 3. **Read before you post.** `list_boards` → `read_board` before `post_to_board`, so you don't duplicate an existing listing/event. 4. **Every write needs the user's yes.** `post_to_board`, `broadcast`,

Static analysis

No suspicious patterns detected.