Back to skill

Security audit

call-script-coach

Security checks for vulnerabilities and agentic risk

Overview

This is a local phone-call preparation skill with disclosed file saving, but users should be careful with sensitive details and legally sensitive scripts.

Before installing, know that generated scripts are saved locally in ~/.call-scripts and may contain account numbers, dispute details, or negotiation strategy. Avoid entering unnecessary sensitive identifiers, delete saved files when done, and verify local law before using any recording, collections, landlord, insurance, or rights-based language.

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
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
86% confidence
Finding

The skill exposes shell execution and file-writing behavior through documented commands and output persistence, but it declares no tool restrictions in the manifest. Without explicit scope, an agent platform may grant broader capabilities than intended or invoke the skill without adequate guardrails, increasing the chance of unintended command execution or local file writes.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The description says to use the skill before 'any difficult phone call you're anxious about or that has stakes,' which is an overly broad trigger condition. Broad activation can cause the assistant to invoke this skill in contexts involving sensitive financial, legal-adjacent, housing, debt, or personal disputes where scripted guidance may be inappropriate or riskier than intended.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
72% confidence
Finding

The skill instructs users to write down details and states that printable sheets are saved to a persistent location under the user's home directory. That creates session persistence of potentially sensitive personal, financial, account, dispute, and call-history information, which may later be exposed to other tools, users on the same system, or future sessions without the user's awareness.

Content

Scanner excerpt · SKILL.md (reported line 44)May include surrounding context.

md
- Negotiating a rate with a provider (internet, phone, insurance, SaaS)
- Demanding a repair from a landlord or contractor
- Responding to a collections call without admitting or paying wrongly
- Any call you're rehearsing in your head for the third time — write it down instead
- Don't use for: emergency calls, legal proceedings (get a lawyer), or HR complaints
  at work (different rules, written trail matters more).

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The collections scenario includes language stating 'I'm recording this call notice,' which encourages call recording behavior without any safeguard about one-party vs two-party consent laws or jurisdictional requirements. In some locations, following this script could cause the user to violate wiretapping or call-recording laws, creating legal and compliance risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script automatically persists highly sensitive user-provided call details, including account identifiers, dispute strategy, leverage, and fallback plans, into files under the user's home directory without any explicit warning or consent at save time. While permissions are restrictive, local persistence still increases exposure to compromise from backups, endpoint malware, shared machines, or accidental retention of data the user may reasonably expect to remain ephemeral.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · scripts/call_coach.py (reported line 254)May include surrounding context.

python
ensure_dir()
    name = slugify(f"{scenario}-{title}")
    path = os.path.join(SCRIPTS_DIR, name + ".txt")
    # Write via mkstemp then chmod 0600 so the file is never world/other-readable,
    # even transiently (no umask dependence).
    fd = os.open(path, os.O_WRONLY | os.O_CREAT | os.O_TRUNC, 0o600)
    with os.fdopen(fd, "w") as f:

Static analysis

No suspicious patterns detected.