Security Audit
agentql-automation
github.com/ComposioHQ/awesome-claude-skillsTrust Assessment
agentql-automation received a trust score of 81/100, placing it in the Mostly Trusted category. This skill has passed most security checks with only minor considerations noted.
SkillShield's automated analysis identified 2 findings: 0 critical, 1 high, 1 medium, and 0 low severity. Key findings include Excessive Tool Permissions via Rube MCP, Untrusted External MCP Dependency.
The analysis covered 4 layers: Manifest Analysis, Static Code Analysis, Dependency Graph, LLM Behavioral Safety. All layers scored 70 or above, reflecting consistent security practices.
Last analyzed on February 17, 2026 (commit 99e2a295). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.
Layer Breakdown
Behavioral Risk Signals
Security Findings2
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| HIGH | Excessive Tool Permissions via Rube MCP The skill instructs the LLM to use highly privileged tools like `RUBE_MULTI_EXECUTE_TOOL` and `RUBE_REMOTE_WORKBENCH` without apparent restrictions. `RUBE_MULTI_EXECUTE_TOOL` allows executing any tool discovered via `RUBE_SEARCH_TOOLS` within the 'agentql' toolkit, and `RUBE_REMOTE_WORKBENCH` with `run_composio_tool()` implies arbitrary Composio tool execution. If the underlying 'agentql' toolkit contains sensitive operations (e.g., data modification, deletion, or access to PII), this skill effectively grants the LLM broad, unrestricted control over those operations. A compromised or misdirected LLM could exploit this to perform unauthorized actions, leading to data loss, corruption, or unauthorized access. Implement fine-grained access control for the LLM's tool usage. Restrict the specific tool slugs or argument patterns that the LLM can pass to `RUBE_MULTI_EXECUTE_TOOL`. For `RUBE_REMOTE_WORKBENCH`, consider if such a broad tool is necessary or if its usage can be limited. Introduce human-in-the-loop approval for sensitive operations or critical tool executions. | LLM | SKILL.md:59 | |
| MEDIUM | Untrusted External MCP Dependency The skill explicitly directs the LLM to connect to an external, unverified Managed Control Plane (MCP) at `https://rube.app/mcp`. Relying on an external MCP introduces a supply chain risk. If `rube.app` were compromised, it could serve malicious tool schemas or execution instructions, leading to the LLM performing harmful actions or exfiltrating data through the MCP. The trustworthiness and security posture of `rube.app` are not established within the skill's context. Verify the trustworthiness and security practices of `rube.app/mcp`. Consider hosting a trusted, audited MCP instance internally or using a well-established and vetted provider. Implement integrity checks (e.g., cryptographic signatures) for tool schemas and execution plans retrieved from the MCP to detect tampering. | LLM | SKILL.md:16 |
Scan History
Embed Code
[](https://skillshield.io/report/697d137582af6760)
Powered by SkillShield