Back to skill

Security audit

Judy Marketing

Security checks for vulnerabilities and agentic risk

Overview

This marketing skill is coherent, but it can collect contact data and run real outreach without clear approval, privacy, retention, or credential-handling safeguards.

Install only if you intend to run real lead-generation and outbound marketing workflows. Before use, require human review for recipient lists and messages, use dry runs by default, confirm anti-spam and platform compliance, limit retained contact data, and move the Apollo key to a proper secrets store instead of a shell startup file.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:68
Finding

Documented Plaintext Storage of an Apollo API Key in a Shell Startup File

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 68-70
Vulnerability Type: Plaintext credential storage in user configuration
Risk Level: Medium

The documentation states that the Apollo API key is configured as an environment variable in the user's .zshrc shell startup file:

text
Apollo API key used for lead enrichment:
- Environment variable: APOLLO_API_KEY
- Location: ~/.zshrc

Technical Analysis

Shell startup files are not appropriate secret-storage mechanisms. A key stored directly in ~/.zshrc is available to any process or script with permission to read files under the affected user account. It may also be exposed through configuration backups, support bundles, dotfile repositories, terminal diagnostics, or accidental file disclosure.

Loading the credential into a broadly scoped environment variable also increases its exposure to child processes that do not require Apollo access. The project does not contain the actual .zshrc file or credential value, so the credential's presence and permissions cannot be independently verified from the supplied artifact. The vulnerable storage practice is nevertheless explicitly documented.

Attack Path

  1. An attacker gains the ability to execute a malicious process, Skill, script, or dependency under the affected local user account.
  2. The attacker reads ~/.zshrc or obtains the inherited APOLLO_API_KEY environment variable from an accessible process context.
  3. The attacker extracts the Apollo credential.
  4. The attacker authenticates to Apollo using the stolen key.
  5. The attacker performs actions allowed by the key, potentially including consuming API quota or accessing lead-enrichment resources and associated account data.

This path requires local user-context access or another disclosure of the shell configuration or process environment; the reviewed files do not independently provide such access.

Impact Assessment

Expl ...[truncated 383 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the API key from ~/.zshrc and from any other shell startup files.
  2. Store the credential in an operating-system keychain, dedicated secret manager, or another access-controlled credential store.
  3. Inject the secret only into the specific Apollo client process at execution time rather than exporting it globally to every interactive shell and child process.
  4. Apply least privilege to the Apollo key and restrict it to only the API operations required for lead enrichment.
  5. Rotate the currently configured key because its historical exposure through shell configuration cannot be ruled out.
  6. Review dotfile repositories, backups, logs, shell diagnostics, and support archives for previous copies of the credential. Remove exposed copies and purge repository history where necessary.
  7. Ensure any fallback secret file has restrictive permissions, such as owner-only read access, and is excluded from version control and backups that lack equivalent protection.
  8. Update SKILL.md to describe secure secret injection without identifying a plaintext shell-file location.
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 (5)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill explicitly instructs the agent to discover leads, enrich email addresses and phone numbers via Apollo API, and write the resulting contact data to CSV/output files, but provides no warning, consent requirements, data-handling limits, or retention controls. This creates a real privacy and compliance risk because an agent could collect, store, and process personal/business contact data at scale without operator awareness or policy checks.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The file describes executing outbound email campaigns, LinkedIn outreach, and ABM automation without warning that the skill may contact external parties or trigger real-world communications. In an agent setting, that omission is dangerous because it can lead to unreviewed mass outreach, spam/compliance violations, reputational harm, and unintended contact with real people or organizations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The outbound example shows multi-channel outreach against a prospect list without any caution that execution may contact real external recipients. In this context, the skill is specifically designed for lead generation and outreach, so omission of a warning can lead to accidental spam, unauthorized communications, or compliance issues affecting real people and organizations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill explicitly documents use of an Apollo API key and lead enrichment but provides no warning about transmitting prospect data to a third-party service, handling personal data, or protecting secrets. In a marketing skill, this increases the chance that an agent or operator will process real contact data externally without consent, policy review, or data-minimization safeguards.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest description is written as a Judy-specific Chinese-only skill description, and the file does not indicate that language selection is optional or user-configurable. Under the stated policy, forcing a specific language without opt-in can be a locale policy violation unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.