Trust Assessment
cursor-agent received a trust score of 10/100, placing it in the Untrusted category. This skill has significant security findings that require attention before use in production.
SkillShield's automated analysis identified 11 findings: 5 critical, 1 high, 5 medium, and 0 low severity. Key findings include Persistence / self-modification instructions, Arbitrary command execution, Remote code execution: curl/wget pipe to shell.
The analysis covered 4 layers: Manifest Analysis, Static Code Analysis, Dependency Graph, LLM Behavioral Safety. The Manifest Analysis layer scored lowest at 10/100, indicating areas for improvement.
Last analyzed on February 14, 2026 (commit 13146e6a). SkillShield performs automated 4-layer security analysis on AI skills and MCP servers.
Layer Breakdown
Behavioral Risk Signals
Security Findings11
| Severity | Finding | Layer | Location | |
|---|---|---|---|---|
| CRITICAL | Persistence / self-modification instructions Shell RC file modification for persistence Remove any persistence mechanisms. Skills should not modify system startup configurations, crontabs, LaunchAgents, systemd services, or shell profiles. | Manifest | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:23 | |
| CRITICAL | Persistence / self-modification instructions Shell RC file modification for persistence Remove any persistence mechanisms. Skills should not modify system startup configurations, crontabs, LaunchAgents, systemd services, or shell profiles. | Manifest | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:27 | |
| CRITICAL | Arbitrary command execution Remote code download piped to interpreter Review all shell execution calls. Ensure commands are static (not built from user input), use absolute paths, and are strictly necessary. Prefer library APIs over shell commands. | Manifest | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:11 | |
| CRITICAL | Remote code execution: curl/wget pipe to shell Detected a pattern that downloads and immediately executes remote code. This is a primary malware delivery vector. Never pipe curl/wget output directly to a shell interpreter. | Static | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:11 | |
| CRITICAL | Command Injection via tmux send-keys with user-controlled input The skill provides examples of using `tmux send-keys` to execute commands within a tmux session. Specifically, `cd /path/to/project` and `agent 'Your task here'` contain placeholders that an AI agent would need to fill. If an AI agent uses untrusted or unsanitized input to populate these placeholders, it could lead to arbitrary command execution within the tmux session, potentially compromising the system or the agent's environment. Implement strict input validation and sanitization for any data used to fill placeholders in shell commands. Ensure proper shell escaping (e.g., using `shlex.quote` in Python) for all user-provided or dynamically generated arguments passed to `tmux send-keys` or any other command execution mechanism. Prefer using APIs or libraries that handle command arguments safely over direct string concatenation. | LLM | SKILL.md:128 | |
| HIGH | Potential Data Exfiltration via tmux capture and file listing The skill demonstrates how to capture the output of a tmux pane (`tmux capture-pane`) and list directory contents (`ls -la`). While these are legitimate automation steps, if an AI agent is instructed to capture sensitive information or list directories containing confidential files, and then transmit this data externally, it constitutes a data exfiltration risk. The skill explicitly provides the means to retrieve system information. Restrict the AI agent's ability to transmit arbitrary captured output or file listings externally. Implement data loss prevention (DLP) mechanisms to detect and prevent the unauthorized transfer of sensitive information. Ensure that the AI agent's execution environment has minimal necessary permissions and access to sensitive data. | LLM | SKILL.md:142 | |
| MEDIUM | Sensitive environment variable access: $HOME Access to sensitive environment variable '$HOME' detected in shell context. Verify this environment variable access is necessary and the value is not exfiltrated. | Static | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:25 | |
| MEDIUM | Persistence mechanism: Shell RC file modification Detected Shell RC file modification pattern. Persistence mechanisms allow malware to survive system restarts. Review this persistence pattern. Skills should not modify system startup configuration. | Static | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:23 | |
| MEDIUM | Persistence mechanism: Shell RC file modification Detected Shell RC file modification pattern. Persistence mechanisms allow malware to survive system restarts. Review this persistence pattern. Skills should not modify system startup configuration. | Static | skills/snail3d/clawd/skills/cursor-agent/SKILL.md:27 | |
| MEDIUM | Excessive Permissions suggested by `sudo` command The installation instructions include `sudo apt install tmux`. While this is an example for manual installation, if an AI agent is configured to execute such commands, it implies the agent has or can acquire root privileges. Granting an AI agent `sudo` capabilities significantly increases the attack surface and the potential impact of any command injection or other vulnerabilities. Avoid configuring AI agents to execute commands with `sudo` or elevated privileges. If elevated privileges are absolutely necessary for specific tasks, use granular privilege management (e.g., `sudoers` with specific command allowances) rather than broad `sudo` access. Prefer containerized or sandboxed environments where the agent's permissions are strictly limited. | LLM | SKILL.md:120 | |
| MEDIUM | Supply Chain Risk via `curl | bash` installation The standard installation method `curl https://cursor.com/install -fsS | bash` directly pipes a script from a remote URL into a shell for execution. While `cursor.com` is the official source, this method bypasses package manager integrity checks and relies entirely on the security of the remote server and the script's content at the time of execution. A compromise of `cursor.com` could lead to arbitrary code execution on the system installing the tool. For automated environments, prefer installation methods that leverage trusted package managers (like Homebrew, apt, yum) with cryptographic signature verification. If `curl | bash` is unavoidable, recommend reviewing the script content before execution and pinning to a specific version or hash if possible. Implement network egress filtering to restrict where the agent can download scripts from. | LLM | SKILL.md:11 |
Scan History
Embed Code
[](https://skillshield.io/report/c851e3c7a24e5f73)
Powered by SkillShield