Back to skill

Security audit

BATNA & ZOPA

Security checks for vulnerabilities and agentic risk

Overview

This negotiation coaching skill is proportionate to its purpose, with a privacy caveat about saving lessons from use.

Review any proposed additions to the reusable skill file and keep them general and anonymous; exclude confidential prices, counterparties, and walk-away positions.

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
Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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)

The skill instructs agents to persist real-use failure patterns without warning users about recording their negotiation details.

Medium
Category
Output Handling
Confidence
86% confidence
Finding

The instruction to add observed entries 'after each real use' and 'paste the actual failure pattern' encourages retaining session-derived information in a reusable skill document without consent or sanitization requirements. In a negotiation skill, this could expose confidential pricing, counterparties, or walk-away positions to later users or repository readers if an agent follows it literally. No automatic write operation or external transmission is specified, so the risk depends on the agent's write permissions and what it records.

Content

No source excerpt is available for this finding.