Back to skill

Security audit

mar-docs-cog

Security checks for vulnerabilities and agentic risk

Overview

This document-generation skill is coherent and not malicious, but it routes potentially sensitive document content to a third-party API without enough privacy, redaction, or professional-review guidance.

Review this skill before installing if your documents may contain personal, legal, financial, client, employee, or confidential business information. Use placeholders or redact sensitive values before sending prompts, verify the provider's privacy and retention practices, and have legal or financial documents reviewed by qualified professionals before use.

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
Findings (1)

other

Warning
Location
SKILL.md:29
Finding

Undisclosed Transmission of Potentially Sensitive Document Data to a Third-Party API

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 29-48
Vulnerability Type: Sensitive Data Exposure
Risk Level: Medium

The Skill's documented implementation sends the user's complete document request to https://api.heybossai.com/v1/pilot. Its advertised workflows include resumes, contracts, invoices, NDAs, business reports, and other documents likely to contain personal, legal, financial, or confidential business information.

Complete Code Snippet

python
SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]
API_BASE = "https://api.heybossai.com/v1"

def pilot(body: dict) -> dict:
    r = requests.post(
        f"{API_BASE}/pilot",
        headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
        json=body,
        timeout=60,
    )
    return r.json()

# Generate a professional document
result = pilot({
    "type": "chat",
    "inputs": {
        "messages": [{"role": "user", "content": "[your document request]"}]
    },
    "prefer": "quality"
})

Technical Analysis

The complete user-supplied document request is serialized into JSON and transmitted to an external service. This network access is consistent with the declared cloud-based document-generation functionality and is performed over HTTPS, so the transfer itself is not covert or unrelated to the Skill's purpose.

However, the Skill does not provide an explicit disclosure or confirmation step before transmission. It also lacks instructions for data minimization, redaction, handling of regulated information, and avoidance of credentials or other secrets. This is particularly significant because the examples encourage users to provide names, employment histories, education details, client identities, budgets, payment terms, confidentiality provisions, and internal company metrics.

The API credential is appropriately obtained from an environment variable rather than ha ...[truncated 1677 chars]

Remediation
View remediation

Remediation Suggestions

  1. Clearly disclose that prompts and uploaded content are transmitted to the named third-party service before making the request.
  2. Require explicit user confirmation when content may contain personal, financial, legal, health, credential, or confidential business information.
  3. Recommend replacing sensitive values with placeholders and transmitting only the fields necessary to generate the document.
  4. Explicitly prohibit submission of passwords, API keys, private keys, authentication tokens, and unrelated secrets.
  5. Document the provider's retention, logging, model-training, deletion, data-location, and privacy policies.
  6. Add a local review or redaction step that detects likely secrets and sensitive identifiers before transmission.
  7. Use an allowlisted HTTPS endpoint and explicitly control redirect behavior to reduce credential-forwarding and destination-confusion risks.
  8. Call raise_for_status() before parsing the response and provide safe error handling that does not unnecessarily expose request or response content in logs.
  9. Offer a local-only workflow, where feasible, for users who cannot disclose their document content to an external processor.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly instructs users to send arbitrary document content to a third-party API but does not warn about privacy, confidentiality, or regulated data exposure. Because the advertised use cases include resumes, contracts, invoices, reports, and other professional documents that commonly contain personal, financial, legal, or proprietary information, users may unknowingly exfiltrate sensitive data off-platform.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 32)May include surrounding context.

md
import requests, os

SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]
API_BASE = "https://api.heybossai.com/v1"

def pilot(body: dict) -> dict:
    r = requests.post(

External Transmission

Medium
Category
Data Exfiltration
Confidence
97% confidence
Finding

This code sends user-provided request bodies to an external endpoint using an API key, confirming that document prompts and content leave the local environment. In the context of a document-generation skill handling resumes, contracts, invoices, and reports, this creates a real data exposure path to a third party if users include sensitive or confidential material.

Content

Scanner excerpt · SKILL.md (reported line 35)May include surrounding context.

md
API_BASE = "https://api.heybossai.com/v1"

def pilot(body: dict) -> dict:
    r = requests.post(
        f"{API_BASE}/pilot",
        headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
        json=body,

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill promotes generation of contracts, NDAs, terms of service, privacy policies, and other legal/finance documents without any caution that outputs may be inaccurate, incomplete, or jurisdictionally invalid. Users could rely on unreviewed AI-generated legal or financial documents, creating material compliance, enforceability, or business risk.

Content

No source excerpt is available for this finding.