Security Audit

composio-search-automation

github.com/ComposioHQ/awesome-claude-skills

AI SkillCommit 27904475d127

TRUSTED

Scanned about 2 months ago

Critical

Immediate action required

High

Priority fixes suggested

Medium

Best practices review

Low

Acknowledged / Tracked

Trust Assessment

composio-search-automation received a trust score of 82/100, 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 Potential Prompt Injection via `use_case` parameter, Dynamic Tool Execution with Broad Scope.

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 20, 2026 (commit 27904475). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.

Layer Breakdown

Manifest Analysis

100%

Static Code Analysis

100%

Dependency Graph

100%

LLM Behavioral Safety

78%

Behavioral Risk Signals

Network Access

1 finding

Shell Execution

2 findings

Dynamic Code

2 findings

Excessive Permissions

1 finding

Security Findings2

Severity	Finding	Layer	Location
HIGH	Dynamic Tool Execution with Broad Scope The skill is designed to dynamically discover and execute tools provided by the Rube MCP via `RUBE_MULTI_EXECUTE_TOOL` and `RUBE_REMOTE_WORKBENCH`. The `tool_slug` and `arguments` for these executions are determined at runtime based on `RUBE_SEARCH_TOOLS` results and agent input. This pattern grants the skill the ability to execute a wide range of operations, potentially including highly privileged or sensitive actions, depending on the tools exposed by the Rube MCP. An attacker could potentially manipulate the agent to execute arbitrary tools available through Rube MCP, leading to unintended actions, data manipulation, or privilege escalation. The skill's documentation explicitly warns to 'Never hardcode tool slugs or arguments without calling `RUBE_SEARCH_TOOLS`', highlighting the dynamic and potentially broad nature of tool access. Implement stricter access controls or a whitelist for allowed `tool_slug` values that `RUBE_MULTI_EXECUTE_TOOL` can execute. Ensure that the Rube MCP itself enforces granular permissions for tools and that the agent is configured with the least privilege necessary. Provide clear warnings to users about the potential for dynamic tool execution to lead to broad access.	LLM	SKILL.md:45
MEDIUM	Potential Prompt Injection via `use_case` parameter The skill instructs the agent to pass user-controlled input, specifically the `use_case` parameter in `RUBE_SEARCH_TOOLS` (e.g., `your specific Composio Search task`), to an external Rube MCP system. If the Rube MCP backend processes this `use_case` using an LLM, a malicious user could craft a `use_case` string to perform prompt injection against the Rube MCP's LLM, potentially manipulating its behavior, leading to unexpected tool suggestions or actions. This represents an indirect prompt injection vector through the skill. The Rube MCP system should sanitize or strictly validate inputs like `use_case` before passing them to an internal LLM. If possible, the skill should guide users to provide structured or constrained inputs for `use_case` to mitigate free-form text injection.	LLM	SKILL.md:37

Scan History

Embed Code

[![SkillShield](https://skillshield.io/api/v1/badge/3cbf7e8c3a320423.svg)](https://skillshield.io/report/3cbf7e8c3a320423)