Security Audit
bigml-automation
github.com/ComposioHQ/awesome-claude-skillsTrust Assessment
bigml-automation received a trust score of 86/100, placing it in the Mostly Trusted category. This skill has passed most security checks with only minor considerations noted.
SkillShield's automated analysis identified 1 finding: 0 critical, 1 high, 0 medium, and 0 low severity. Key findings include Excessive Permissions via Generic Tool Execution.
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 Findings1
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| HIGH | Excessive Permissions via Generic Tool Execution The skill's manifest declares a broad dependency on the entire 'rube' MCP ecosystem (`mcp: ['rube']`). The documentation then instructs on using `RUBE_SEARCH_TOOLS` to discover available tools and `RUBE_MULTI_EXECUTE_TOOL` to execute any tool slug returned by the search. While the skill is advertised for 'Bigml automation' and examples for `RUBE_SEARCH_TOOLS` suggest querying for 'Bigml operations', there are no explicit constraints in the skill's definition or usage examples that prevent `RUBE_SEARCH_TOOLS` from returning tools outside the Bigml domain, or `RUBE_MULTI_EXECUTE_TOOL` from executing them. This grants the skill broad execution capabilities beyond its stated purpose, potentially allowing it to interact with arbitrary systems or resources if the Rube MCP exposes such tools (e.g., filesystem access, network requests to arbitrary domains, environment variable manipulation). The `RUBE_REMOTE_WORKBENCH` with `run_composio_tool()` further indicates a generic tool execution capability. 1. **Restrict Tool Discovery**: Modify the skill's logic to explicitly filter or constrain the `queries` parameter in `RUBE_SEARCH_TOOLS` to only return tools related to 'bigml'. 2. **Validate Tool Execution**: Implement checks to ensure that any `tool_slug` passed to `RUBE_MULTI_EXECUTE_TOOL` or `RUBE_REMOTE_WORKBENCH` belongs to the 'bigml' toolkit or a predefined, safe list of allowed tools. 3. **Granular Permissions**: If the platform supports it, specify more granular permissions in the manifest (e.g., `mcp: ['rube:bigml']`) instead of a broad `mcp: ['rube']` to adhere to the principle of least privilege. 4. **Limit Generic Execution**: Carefully evaluate the necessity and scope of `RUBE_REMOTE_WORKBENCH` usage if its `run_composio_tool()` function allows arbitrary tool execution. | LLM | SKILL.md:43 |
Scan History
Embed Code
[](https://skillshield.io/report/efb86dc2c320de63)
Powered by SkillShield