Back to skill

Security audit

Generate Estate Planning Law Firm Client Education Handout

Security checks for vulnerabilities and agentic risk

Overview

The skill appears intended for legitimate legal client-education drafting, but it grants broad shell and file-read access that is not explained by that purpose.

Review this skill before installing. Its drafting purpose is understandable, but it should not need broad Bash or unrestricted file-read access; install only if you are comfortable with those permissions or can remove/narrow them. Any output should be reviewed by a qualified attorney before client use.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:4
Finding

Excessive Shell and Filesystem Permissions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 4
Vulnerability Type: Excessive tool permissions violating least privilege
Risk Level: Medium

Complete Code Snippet:

yaml
allowed-tools: Bash, Read

Technical Analysis

The skill is intended to generate an estate-planning client education handout. Its documented workflow involves drafting, refining, and optionally enriching content with chat and image-generation capabilities. It does not document a legitimate need to execute arbitrary shell commands or read unrestricted local files.

Granting Bash and Read therefore exposes capabilities beyond the skill's stated operational requirements. In an environment where skill instructions, user prompts, or referenced content can influence tool use, these permissions could be abused to execute local commands or access files unrelated to the handout-generation task.

The reviewed file does not itself contain malicious commands, instructions to read sensitive files, or an exfiltration mechanism. The vulnerability is the unnecessary privilege exposure and resulting attack surface.

Attack Path

  1. The skill is loaded with the declared Bash and Read permissions.
  2. An attacker supplies adversarial task content or untrusted material designed to influence the agent's tool selection.
  3. The agent is induced to invoke Read against files outside the legitimate task scope or to run commands through Bash.
  4. Local information may be exposed to the active session, or shell commands may execute with the privileges of the hosting agent process.
  5. Further impact depends on host permissions and available integrations; no persistence or external exfiltration path is defined in the reviewed skill.

Impact Assessment

Successful exploitation could provide access to files readable by the agent process and permit command execution under that process's operating-system identity. The affected scope may ...[truncated 383 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove Bash and Read from allowed-tools because they are not required by the documented handout-generation workflow.
  2. Allow only narrowly scoped capabilities necessary for the task, such as approved chat and image-generation tools.
  3. If document input is required, replace unrestricted filesystem access with a constrained reader limited to files explicitly selected or uploaded by the user.
  4. If command execution later becomes necessary, expose specific allowlisted operations instead of a general-purpose shell.
  5. Enforce runtime filesystem boundaries, command allowlists, and explicit user confirmation for access to sensitive resources.
  6. Keep human review mandatory before generated material is delivered to clients or treated as legal advice.
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The description is broad and generic enough that an orchestrator or user may invoke this skill in situations outside a narrowly defined estate-planning handout task. That can cause unintended use on legal-content generation requests without sufficient scope checks, increasing the chance of overbroad drafting or reliance on incomplete legal context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The 'Use When' guidance is underspecified and does not define scope constraints, exclusions, or minimum required context before activation. In a legal domain, that ambiguity can lead to the skill being used for quasi-legal drafting or client-facing materials beyond simple education, which raises compliance and accuracy risk.

Content

No source excerpt is available for this finding.