Back to skill

Security audit

TestTau

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent TestTau testing helper, but it can automatically route email and webhook data to a third-party service and exposes wipe/replay actions without strong user-control guidance.

Install only if you intend agents to use TestTau for non-production QA or CI data. Avoid public inboxes and hooks for secrets, PII, account recovery, or production workflows; prefer private endpoints with bearer tokens, and require explicit user confirmation before replaying requests or using any wipe endpoint.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/api.md (reported line 26)May include surrounding context.

md
- `GET /message/<messageId>` - parsed message JSON.
- `GET /message/<messageId>/json` - parsed message JSON.
- `GET /message/<messageId>/raw` - raw `.eml`.
- `DELETE /message/<messageId>` - delete one message.
- `DELETE /all` - wipe inbox.

## Hooks

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

DELETE /all enables bulk destruction of inbox contents, which is materially riskier than deleting a single identified resource. In an agent skill context, exposing a one-step wipe operation without any safety guidance, confirmation requirement, or scoping notes increases the chance of accidental or induced destructive actions that can erase forensic or test evidence.

Content

Scanner excerpt · references/api.md (reported line 27)May include surrounding context.

md
- `GET /message/<messageId>/json` - parsed message JSON.
- `GET /message/<messageId>/raw` - raw `.eml`.
- `DELETE /message/<messageId>` - delete one message.
- `DELETE /all` - wipe inbox.

## Hooks

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · references/api.md (reported line 64)May include surrounding context.

md
- `PUT /config` - set response behavior or JSON Schema.
- `POST /replay/<requestId>?target=<url>` - replay capture.
- `GET /assert?since=<epochMs>&min_count=1` - CI assertion gate.
- `DELETE /request/<requestId>` - delete one capture.
- `DELETE /all` - wipe hook.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% confidence
Finding

DELETE /all for hooks permits complete wiping of captured requests, creating a destructive primitive that an autonomous agent could misuse if prompted or if context is ambiguous. In a testing and inspection skill, this is more dangerous because captures may contain evidence needed for QA, debugging, or incident review, and bulk deletion can remove all of it at once.

Content

Scanner excerpt · references/api.md (reported line 65)May include surrounding context.

md
- `POST /replay/<requestId>?target=<url>` - replay capture.
- `GET /assert?since=<epochMs>&min_count=1` - CI assertion gate.
- `DELETE /request/<requestId>` - delete one capture.
- `DELETE /all` - wipe hook.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

  1. Wait for the message:
bash
curl -fsS "https://mail.testtau.com/i/agent-login-7k3p/api/wait?timeout=15000&subject=Verify"
  1. Read parsed JSON:

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The default prompt encourages broad use of the skill whenever an agent needs disposable inboxes or webhooks, but it does not define narrow activation conditions, authorization checks, or user-confirmation requirements. In an agent environment, this can cause over-invocation and unintended data routing to third-party disposable infrastructure, especially when handling emails, API-key protected test inboxes, or webhook captures during ambiguous tasks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Enabling implicit invocation without contextual restrictions allows the agent to call this skill automatically based on loose relevance rather than explicit user intent. Because the skill creates disposable mailboxes and captures webhook/request data, automatic invocation increases the chance of unintended external data exposure, accidental interception of test or semi-sensitive communications, and expansion of the agent's effective action surface.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

This markdown file describes DELETE /message/<messageId> and DELETE /all operations, including a full inbox wipe, but provides no warning that these actions are destructive and may permanently remove user data. For markdown files, safety-affecting behaviors should be disclosed so users understand the impact before using the skill or API.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The file lists DELETE /request/<requestId> and DELETE /all for hooks, including wiping all captured requests, but does not warn users about data loss or irreversible effects. In markdown documentation, destructive behaviors that affect stored user or system data should be clearly called out.

Content

No source excerpt is available for this finding.