Back to skill

Security audit

DoctorClaw Invoice Tracker

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed invoice-tracking workflow that reads configured invoice data, drafts reminders, and only sends payment emails after user approval.

Before installing, configure exactly which invoice source the skill may read, where digests may be delivered, and whether weekly scheduling is enabled. Do not grant email or accounting access unless you want the agent to draft and, after explicit approval, send reminders or update invoice records.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill invites broad use with wording like running weekly or whenever needed, but does not define a narrow invocation boundary or explicit preconditions. In an agent environment, this can cause accidental activation on loosely related finance requests and trigger reading sensitive invoice data or preparing customer-facing actions unexpectedly.

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 · SKILL.md (reported line 84)May include surrounding context.

md
- **Third reminder (15-30 days overdue):** Firm. "Invoice #X is now [X] days past due. Please arrange payment at your earliest convenience."
- **Escalation (30+ days):** Serious. "This is a final notice regarding Invoice #X. If we don't receive payment by [date], we may need to discuss next steps."

Match the configured reminder style. Mark all as DRAFT — never send without approval.

### Step 4: Compile Invoice Digest
Format the digest:

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The example trigger 'Check my invoices' is generic enough to overlap with normal conversational requests about invoices. In a multi-skill agent, this increases the chance of unintended routing into this skill, exposing financial data summaries or initiating reminder drafting without the user intending this exact workflow.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.