ClawHub
Back to skill

Security audit

team-discuss

Security checks for vulnerabilities and agentic risk

Overview

This skill is a transparent multi-agent discussion helper, with expected local transcript storage and optional sub-agent use that users should handle carefully.

Install this only if you are comfortable with discussion content being saved in local files and, when configured, sent to participating sub-agents. Avoid putting secrets, private customer data, legal or medical details, or other regulated information into discussions unless you have reviewed the storage path, permissions, and agent trust boundaries.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill describes shared file-based persistence without warning that discussion content will be stored on disk, which can surprise users and lead to unintended retention of sensitive internal deliberations. In collaborative agent workflows, persisted transcripts may be accessible longer than expected or to other local processes/users depending on filesystem permissions and storage location.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The documentation highlights real sub-agent invocation via sessions_spawn but does not disclose that discussion prompts, prior messages, and possibly sensitive context are transmitted to other agents. This creates a data-handling transparency gap: users may unknowingly send confidential information to downstream components with different trust boundaries, retention, or logging behavior.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The skill advertises applicability to essentially any multi-perspective topic, which makes its invocation boundary overly broad and increases the chance it will be selected for sensitive, regulated, or high-risk decisions without appropriate safeguards. In an agent ecosystem, vague scope can cause inappropriate routing and accidental exposure of confidential prompts or decision authority to a tool not designed for those contexts.

Static analysis

No suspicious patterns detected.