Back to skill

Security audit

TOSR Test Skill

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed lifecycle test skill that describes live create, update, inspect, and delete operations for a specific ephemeral test skill, with no bundled code or hidden behavior.

Install or use this only in an authorized ClawHub test context. Treat the described API actions as live mutations, keep them scoped to the declared test slug, and avoid running lifecycle cleanup against production or user-owned skills unless you explicitly intend that.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • 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 (3)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
89% confidence
Finding

The skill advertises a DELETE operation on /api/v1/skills/{slug}, which is a high-risk action because the slug parameter controls which resource is destroyed. Without strict validation, scoping, and confirmation, an agent could delete the wrong skill or be induced to target arbitrary resources, making this materially dangerous despite the stated test purpose.

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

md
1. **Publish** — Creates a new skill via POST /api/v1/skills with multipart form data
2. **Inspect** — Retrieves skill metadata via GET /api/v1/skills/{slug}
3. **Update** — Publishes a new version of an existing skill
4. **Delete** — Removes the skill via DELETE /api/v1/skills/{slug}

## Notes

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill explicitly describes publishing, updating, and deleting skills against a real API, including a destructive DELETE operation, but it does not include a clear user-facing warning, authorization requirement, or confirmation step about those effects. Even if intended for integration testing, documenting destructive capability without guardrails increases the chance an agent or operator could invoke it inappropriately against real resources.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

The statement that the skill will be 'automatically deleted after the test completes' indicates autonomous destructive behavior without showing any approval gate or runtime validation. In context this is cleanup logic for an ephemeral test skill, so the danger is limited, but unattended deletion behavior can still be risky if targeting or environment checks fail.

Content

Scanner excerpt · SKILL.md (reported line 28)May include surrounding context.

md
## Notes

This skill is ephemeral and will be automatically deleted after the test completes.
If you see this skill listed on clawhub, it means a test run failed to clean up properly.