Back to skill

Security audit

Cli Memo

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it can activate for unrelated questions and permits unreviewed CRM changes, though the scan reports no executable files or install mechanism concerns.

Review the skill's activation rules and CRM permissions before installing. Consider whether you need a memory skill that can make CRM changes without review, and whether those changes can be limited to explicit requests and checked before they take effect.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (3)

The activation instruction includes any question about a person or project without requiring a connection to the owner's stored context.

Medium
Category
Trigger Abuse
Confidence
88% confidence
Finding

Although the description explains that memo searches the owner's records, its explicit invocation condition says to use it whenever asked about a person or project or what was agreed. This can also match general-knowledge questions that do not call for searching private messages, mail, or notes.

Content

No source excerpt is available for this finding.

The skill expands knowledge retrieval and memory recording into autonomous CRM management, including server-side changes.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

A skill advertised for retrieval and recording memories explicitly permits immediate deal creation, movement, closure, and pipeline management without review, and provides a route to server-side CRM writes. This expands the agent's authority beyond the advertised task and bypasses the proposal-and-owner-review model used for other consequential records, allowing mistaken or unrequested changes to business state.

Content

No source excerpt is available for this finding.

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 · topics/proposals.md (reported line 41)May include surrounding context.

md
One person can have a Telegram account, a MAX account, a CRM contact and an email address. Each is an identity,
and memo joins identities into one person.

**What joins people without asking.**

- The same email address, in every account, work or personal, when no more than two people have it and it is not a
  role address (`info@`, `support@` …): a role address only proposes, and a no-reply address does nothing.