T09 · Insecure Skill Coding Practices
Warning
- Location
scripts/status.sh:5- Finding
Hardcoded Reusable Supabase Anonymous API Token
- Content
View full analysis
- Remediation
View remediation
Security audit
Clawbuddy Litekit
Security checks for vulnerabilities and agentic risk
Overview
The skill is mostly coherent with its dashboard purpose, but it asks users to delegate an OpenClaw API key to a hosted cloud proxy and includes an under-disclosed remote status check.
Review before installing or using setup. Only provide a dedicated, least-privilege OpenClaw Gateway key, understand that it will be stored and used by hosted Lovable Cloud infrastructure, and revoke it when no longer needed. Running the status script contacts a Supabase-hosted endpoint using a bundled anonymous token.
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 (2)
T09 · Insecure Skill Coding Practices
Warning
- Location
SKILL.md:32- Finding
Gateway Credential Authority Delegated to an External Hosted Proxy
- Content
View full analysis
- Remediation
View remediation
Vulnerability Patterns
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)
Tp4
High
- Category
- MCP Tool Poisoning
- Confidence
- 97% confidence
- Finding
The documented behavior says the skill opens a visual dashboard and setup flow, but the analyzed behavior reportedly performs backend status/configuration checks and contacts an undisclosed remote Supabase endpoint using hardcoded connection details. This mismatch is dangerous because it prevents informed consent, hides external data flows, and can be used to exfiltrate metadata or interact with remote infrastructure under the guise of a harmless UI action.
- Content
Undeclared Tool Scope
Medium
- Category
- MCP Least Privilege
- Confidence
- 88% confidence
- Finding
The skill advertises shell-based quick actions but does not declare any explicit tool scope or allowed-tools boundary. That creates an authorization and review gap: an agent may invoke shell commands without the user or platform having a clearly constrained permission model, increasing the chance of unintended local command execution.
- Content
Description-Behavior Mismatch
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The manifest describes a skill for opening a mission-control dashboard and visually inspecting agent and gateway information. This script instead performs a backend HTTP call to an
openclaw-statusfunction to determine whether gateway secrets are configured, which is a setup/health-check operation rather than opening or presenting the dashboard itself.- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The script silently sends authentication material in both
apikeyandAuthorizationheaders to a remote Supabase function without any user disclosure or consent prompt. Even though the embedded token appears to be an anonymous key, transmitting credentials from a local agent skill to a remote service creates an undisclosed trust boundary and can expose metadata about skill execution while normalizing hardcoded secret-like values in distributed scripts.- Content