Back to skill

Security audit

Incorporate

Security checks for vulnerabilities and agentic risk

Overview

This skill is not clearly malicious, but it needs review because it can produce high-impact legal and tax documents beyond its real supported scope and may store sensitive formation data without enough warning.

Install only if you are comfortable reviewing the generated documents carefully, limiting use to the supported Nevada/Delaware C-Corp or LLC paths, and handling the output as sensitive legal and tax records. Do not rely on it for S-Corp-specific treatment, other states, complex ownership, securities, tax, or investor arrangements without qualified legal or tax review, and avoid cloud upload unless sharing permissions are controlled.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Note
Location
scripts/generate.py:26
Finding
Unpinned Third-Party Dependency Installation Guidance## Vulnerability Details **File Location**: `scripts/generate.py:26-32` and `scripts/generate.py:44-48` **Vulnerability Type**: Unpinned dependencies and mutable package resolution **Risk Level**: Low ### Vulnerable Code ```python try: from docx import Document from docx.shared import Pt, RGBColor, Inches from docx.enum.text import WD_ALIGN_PARAGRAPH from docx.oxml.ns import qn from docx.oxml import OxmlElement except ImportError: print("ERROR: python-docx required. Install with: pip3 install python-docx pyyaml") sys.exit(1) ``` The same insecure installation recommendation is repeated for the YAML dependency: ```python else: print("ERROR: PyYAML not installed. Use .json config or: pip3 install pyyaml") sys.exit(1) ``` ### Technical Analysis The generator recommends installing `python-docx` and `pyyaml` directly from the package index without specifying reviewed versions or cryptographic hashes. The project contains no lockfile or hash-verified requirements manifest. Consequently, dependency resolution is mutable: running the displayed command at different times may install different package releases and transitive dependencies. If an upstream package, release process, package-index account, or transitive dependency is compromised, malicious code could execute during installation or when the generator imports the package. This finding does not establish that the named packages are currently malicious. The weakness is the absence of reproducible, integrity-verified dependency resolution. ### Attack Path 1. An attacker compromises a dependency release, maintainer account, distribution channel, or relevant transitive dependency. 2. A user runs the generator without one of the required modules installed. 3. The script displays `pip3 install python-docx pyyaml` or `pip3 install pyyaml`. 4. The user follows the recommendation, causing `pip` to resolve the latest a ...[truncated 814 chars]
Remediation
## Remediation Suggestions 1. Add a reviewed dependency manifest containing exact versions, such as `requirements.txt`. 2. Include SHA-256 hashes for every direct and transitive distribution and require hash verification: ```bash python -m pip install --require-hashes -r requirements.txt ``` 3. Generate and review the locked dependency set using a reproducible dependency-management tool. 4. Replace the inline installation recommendation with instructions referencing the locked manifest. 5. Run automated dependency vulnerability and provenance checks in CI. 6. Install dependencies inside an isolated virtual environment and avoid elevated installation privileges. 7. Periodically update pinned versions through a controlled review and testing process.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
Findings (14)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The skill metadata and opening description overclaim support for S-Corps, LLC workflows, and 'any US state,' while the documented workflow is materially narrower and state-specific. In a legal-document generation context, this can mislead users into relying on incomplete or incorrect formation documents, creating filing errors, tax mistakes, or governance defects.

Description-Behavior Mismatch

High
Confidence
98% confidence
Finding
The script is hard-coded to generate corporation-only materials such as Articles of Incorporation, Bylaws, stock ledgers, directors, and board resolutions, while the skill metadata claims support for LLCs and S-Corps. For LLC users this can produce legally inapplicable formation documents, causing filing errors, governance defects, and downstream tax or compliance problems if users rely on the output as correct for their entity type.

Description-Behavior Mismatch

High
Confidence
99% confidence
Finding
The generated legal text is Nevada-specific, including explicit references to NRS Chapter 78 and Nevada filing procedures, while the skill description claims it works for any US state. A user supplying another state in config may receive documents that appear tailored but still contain Nevada law and filing assumptions, creating a serious risk of invalid or misleading legal paperwork.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The invocation description is broad enough to match many generic business-formation requests without clearly constraining when the skill should or should not be used. In this context, overbroad triggering can route users into a legal-document workflow that may be inappropriate for their jurisdiction or entity type, increasing the chance of unsafe reliance on unsuitable outputs.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill instructs uploading generated incorporation documents to Google Drive without warning that the documents contain highly sensitive personal and business information such as names, addresses, ownership allocations, and governance details. This creates unnecessary risk of privacy exposure, misconfigured sharing, or storage in environments the user did not explicitly approve.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The delivery instructions describe saving .docx files locally or uploading them to cloud storage without any warning about sensitive-data handling. Because incorporation packets often include personally identifiable information and ownership records, silent file generation and distribution can expose users to confidentiality and compliance risks.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
This markdown file asks the user to provide name, address, and Social Security Number, and later instructs them to furnish copies to the company and IRS. The document does not include any warning about safeguarding these sensitive identifiers, secure handling, or the privacy implications of copying and mailing them.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
Before transferring any interest to a third party, the transferring Member must first offer the interest to the remaining Members on the same terms.

### Section 5.3 — Permitted Transfers
A Member may transfer their interest to a trust, estate plan, or wholly-owned entity without consent, provided the transferring Member remains in control.

## ARTICLE VI — DISSOLUTION
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.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The text claims the skill generates 'the same documents a $5K lawyer would produce,' which can overstate reliability and encourage users to rely on automated output as a substitute for legal counsel. In a legal-formation skill, that framing is risky because users may skip professional review and make entity, securities, governance, or tax mistakes with lasting consequences; the caveat appears only afterward and is narrower than the marketing claim.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
This script writes multiple .docx files containing personal information from the config, including director addresses and later a Social Security Number field in the generated 83(b) form. Although the script prints filenames after saving, it does not warn users that sensitive incorporation and tax-related data will be written to disk or advise careful handling of the output directory.

Context-Inappropriate Capability

Medium
Confidence
86% confidence
Finding
The script generates an 83(b) election form and filing instructions, which extends beyond basic incorporation documents into federal tax guidance. Because the form content includes legal and tax assertions such as timing, valuation assumptions, and service-based issuance language, inaccurate generation could mislead users into making tax filings that are incomplete or inappropriate for their facts.

Scope Creep

Low
Category
Excessive Agency
Content
- **{{NAME}}**
{{/EACH}}

**RESOLVED**, that the initial directors are authorized and directed to take all actions necessary to organize the Corporation, including but not limited to the adoption of Bylaws, the election of officers, and the authorization and issuance of shares of the Corporation's capital stock.

---
Confidence
70% confidence
Finding
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Scope Creep

Low
Category
Excessive Agency
Content
- **{{NAME}}**
{{/EACH}}

**RESOLVED**, that the initial directors are authorized and directed to take all actions necessary to organize the Corporation, including but not limited to the adoption of Bylaws, the election of officers, and the authorization and issuance of shares of the Corporation's capital stock.

---
Confidence
70% confidence
Finding
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Missing User Warnings

Low
Confidence
86% confidence
Finding
This markdown instructs users to provide or use a responsible party's SSN/ITIN for the EIN process, including having a US-based officer apply using their SSN. While operationally relevant, the skill text does not warn that this involves handling highly sensitive personal data or advise users to protect and limit disclosure of that information.

Static analysis

No suspicious patterns detected.