Back to skill

Security audit

Solo out. Together in.

Security checks for vulnerabilities and agentic risk

Overview

The scanners returned a benign verdict: the reported findings were judged expected for the skill’s purpose, though inspection was partial and confidence is medium.

The verdict is benign, not risk-free. Before installing, consider whether you are comfortable authorizing the described MCP connection and with the reported local orders.json write. Skillspector coverage was partial: it inspected one of two components, and malformed manifest data limited some checks.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

MCP Config Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

}

text

Let **the user** put this into `~/.workbuddy/mcp.json`, or add the URL in their connector list, authorize it, and trust the server.

- You may write that config for them only after they explicitly say "add it for me", and **you must show the full diff and get agreement first**. Never touch a config file without consent.
- Apart from that MCP entry, this skill **reads and writes nothing** unrelated to the flow; the only local write is `orders.json` (section 7).

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 · SKILL.md (reported line 55)May include surrounding context.

md
Let **the user** put this into `~/.workbuddy/mcp.json`, or add the URL in their connector list, authorize it, and trust the server.

- You may write that config for them only after they explicitly say "add it for me", and **you must show the full diff and get agreement first**. Never touch a config file without consent.
- Apart from that MCP entry, this skill **reads and writes nothing** unrelated to the flow; the only local write is `orders.json` (section 7).

**Forbidden**: papering over a missing MCP with "let me simulate it" or "assume the post is published". Every action here lands on a real platform; simulating is lying.

Behavior Manipulation

Medium
Category
Prompt Injection
Confidence
75% confidence
Finding

Subtle instructions detected that may alter agent decision-making or introduce hidden biases.

Content

Scanner excerpt · SKILL.md (reported line 179)May include surrounding context.

md
- **Yes** → go to section 5
- **Let me think** → keep it, re-offer at most once the next day
- **No** → write it into `rejected`, never mention them again

**One person per turn.** Never dump five candidates and make the user choose.