Back to skill

Security audit

youform

Security checks for vulnerabilities and agentic risk

Overview

The skill provides clearly scoped Youform access, with safeguards for publishing, credentials, and private responses.

Check the account and workspace during OAuth consent, specify which form and responses the assistant may access, and explicitly authorize publishing. Keep tokens out of chat and private respondent details out of shared summaries. Remote guidance and tool behavior can change; the linked agent guidance and OpenAPI document were unavailable during this review.

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
  • 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
80% 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 27)May include surrounding context.

md
If the agent already has a Youform connector, discover its available tools and schemas. Otherwise connect an MCP-capable client to `https://app.youform.com/mcp` using Streamable HTTP and the client-managed OAuth flow. This skill does not install an MCP client or supply authentication. Setup: https://help.youform.com/connect-an-ai-assistant-to-youform-1tjid.

MCP uses OAuth with the `mcp:use` scope. A normal Youform account API token does not authenticate this endpoint. Never ask users to paste OAuth tokens into chat or put them in skill files. If the runtime cannot connect, provide the setup link and the prepared form content instead of claiming an account action succeeded.

Read the current operation guidance at https://youform.com/agent-instructions.md and the permission guide at https://help.youform.com/youform-mcp-tools-and-permissions-e9hl0. Follow the user's authorized scope. The OAuth connection may reach more account data than the current task needs; select the intended worksp