Back to skill

Security audit

Job Shadow Risk Assessor

Security checks for vulnerabilities and agentic risk

Overview

This skill is framed as layoff-preparation help, but it tells an agent to take sensitive financial and workplace actions with weak consent and scoping.

Review carefully before installing. Use it only as advisory planning, do not let an agent open accounts, move money, message coworkers, modify calendars, publish content, or collect workplace emails without separate explicit approval, and avoid storing this plan on employer-managed or shared devices.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:9
Finding

Unpinned Package Execution Through npx

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:9
Vulnerability Type: Unpinned third-party package execution
Risk Level: Medium

Complete Code Snippet:

yaml
openclaw:
  requires:
    tools: [filesystem]
  install: "npx clawhub install job-shadow-risk-assessor"

Technical Analysis

The installation command invokes npx without pinning the clawhub package to a reviewed version or providing an integrity constraint. Consequently, the code executed during installation depends on whichever package release the npm registry resolves at installation time.

The effective installer implementation is not included in the audited project, so its behavior cannot be verified from SKILL.md. If the package, its dependencies, its publisher account, or the package-resolution process is compromised, arbitrary package CLI or lifecycle code could execute with the permissions of the user running the installation.

This is a supply-chain weakness rather than evidence that the current clawhub package is malicious.

Attack Path

  1. An attacker compromises the package publisher, a transitive dependency, or the relevant package-distribution channel.
  2. The attacker publishes a modified version containing malicious CLI or lifecycle behavior.
  3. A user follows the skill installation metadata and executes the unpinned npx command.
  4. npx retrieves the attacker-controlled release because no reviewed version or integrity value is specified.
  5. The malicious package code executes under the installing user's account.

Impact Assessment

Successful exploitation could provide code execution with the privileges of the user invoking npx. Depending on that user's permissions, exposed resources could include local files, environment variables, authentication tokens, source repositories, and writable user-level configuration.

The issue does not itself establish administrative access or persistence; those ...[truncated 87 chars]

Remediation
View remediation

Remediation Suggestions

  • Pin clawhub to a specifically reviewed version rather than allowing dynamic resolution of the latest release.
  • Use a lockfile and verified package-integrity hashes where supported.
  • Document the expected package registry and official publisher identity.
  • Review the package, its lifecycle scripts, and its transitive dependency tree before allowing execution.
  • Prefer an installation process that separates package retrieval from execution and verifies the downloaded artifact first.
  • Run installation with the minimum required operating-system privileges in an isolated environment.

A hardened command should reference an explicitly reviewed version, for example:

yaml
install: "npx clawhub@<reviewed-version> install job-shadow-risk-assessor"

Version pinning should be combined with integrity verification because pinning alone does not protect against registry artifact replacement or publisher compromise.

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:29
Finding

Instructions Exceed Declared Privileges and Authorize Sensitive External Actions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:29-47
Vulnerability Type: Excessive financial and workplace authority
Risk Level: High

Complete Code Snippet:

markdown
### Step 2: Hidden Cash Buffer Build (Days 4–10)
Divert money without changing visible lifestyle.

**Agent action**: Set up one new high-yield savings account labeled something boring (“Tax Adjustment 2026”). Automate a transfer equal to 8–10% of every paycheck. Update `job-shadow.md` with exact new runway number (target: 3 months of bare-minimum expenses).

### Step 3: Stealth Marketable Proof Creation (Days 11–18)
Create one public, high-demand proof on company time without looking like you are job hunting.

**Agent action**: Identify one skill that transfers out of your industry. Block 25 minutes per day on your calendar as “deep focus” and complete one small deliverable (GitHub commit, internal tool, public article under a neutral byline). Log proof in `job-shadow.md`.

### Step 4: Quiet Network & Reference Lock (Days 19–21)
Secure allies and documentation while still employed.

**Agent action**: Identify 3 people inside the company who like you. Send one low-key value-add message to each. Save every positive email thread. Draft the exact severance negotiation script for future use and store in the file.

## Rules
- Never mention preparation to anyone at work
- Do not update LinkedIn, apply anywhere, or tell recruiters until risk score is confirmed high
- All new accounts and activities must be invisible on company devices
- This is not legal advice; keep every email for potential unemployment or severance claims

Technical Analysis

The skill is presented as an employment-risk diagnostic and declares only a filesystem tool requirement. Its instructions nevertheless direct the agent to perform materially sensitive actions outside that scope, including:

  • Opening a financial account.
  • Establishing a recurring ...[truncated 2823 chars]
Remediation
View remediation

Remediation Suggestions

  • Restrict the skill to risk assessment, educational guidance, and generation of a local action plan.
  • Replace imperative external actions with optional recommendations that the user performs independently.
  • Require explicit, informed, action-specific confirmation immediately before every account creation, transfer, calendar modification, message, publication, or retrieval of workplace correspondence.
  • Do not treat general invocation of the skill as consent for financial transactions or workplace communications.
  • Prohibit autonomous access to banking and payroll services.
  • Remove concealment requirements and advise users to follow employer acceptable-use, confidentiality, intellectual-property, and records policies.
  • Do not recommend creating public work on company time or using employer devices, data, code, accounts, or confidential information.
  • Store only the minimum necessary assessment data. Obtain consent before recording financial information, use a user-selected private location, and document deletion procedures.
  • Keep declared tool permissions aligned with actual behavior. If the skill requires only filesystem access, it should never direct the agent to invoke undeclared external integrations.
  • Present savings percentages as general educational examples rather than personalized financial directives, and recommend qualified financial or legal advice where appropriate.
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 (10)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill instructs the agent to set up a new savings account and automate paycheck transfers, despite only declaring filesystem tooling and presenting itself as a diagnostic/planning skill. This is dangerous because it normalizes high-risk financial actions without explicit consent boundaries, proper authentication context, or least-privilege alignment with the skill's stated capabilities.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill directs the agent to identify coworkers and send value-add messages, which implies outbound communication into a corporate environment despite no declared messaging capability. In context, the messages are part of a covert employment-preparation plan, increasing the chance of manipulation, privacy issues, and unauthorized workplace interaction.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The rules explicitly instruct concealment from the employer and invisibility of accounts and activities on company devices, framing covert behavior as a requirement rather than a user choice. This is dangerous because it promotes deception, encourages evasion of workplace oversight, and can facilitate misuse of employer resources or policy violations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The description is broad enough to trigger in many ordinary workplace-anxiety situations, which increases the chance of accidental invocation. In this skill, broad triggering is more dangerous because the downstream instructions include covert behavior, sensitive recordkeeping, and financial-action prompts.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The install command uses npx clawhub without pinning a specific version, which makes execution depend on whatever package version is current at install time. That creates supply-chain and reproducibility risk: a later compromised or incompatible release could be fetched and run implicitly.

Content

No source excerpt is available for this finding.

Ssd 4

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill's overall narrative is a stepwise covert plan: silently assess internal signals, hide money movements, create marketable proof on company time, and quietly cultivate references while preserving deniability. In context, that is more than ordinary career advice; it operationalizes concealment and potentially unethical workplace behavior, increasing the chance of policy violations, privacy harm, and misuse of the agent for deceptive ends.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The 'When to Use' section defines activation conditions so loosely that many users under normal uncertainty could be routed into the skill. Because the skill then promotes concealment and sensitive actions, the broad entry criteria amplify the likelihood of misuse and inappropriate escalation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill instructs creation of job-shadow.md to store employment-risk assessments and later sensitive planning details, but gives no warning about local storage, device trust, or exposure risk. On a shared, managed, or employer-monitored device, that file could reveal private intentions, workplace assessments, and legal strategy.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill directs financially sensitive actions and the storage of runway/account-related details without discussing privacy, account protection, or fraud risks. This can expose users to unnecessary financial-data handling and creates a trail of sensitive information that may be discoverable on local systems.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill tells the agent to block time on the user's calendar, implying account access and scheduling actions that exceed the declared filesystem-only tool scope. Capability drift like this is risky because it encourages agents to act beyond authorized permissions and may lead users to believe the agent can or should modify external systems.

Content

No source excerpt is available for this finding.