← Back to Skills Marketplace
146
Downloads
0
Stars
0
Active Installs
8
Versions
Install in OpenClaw
/install cmic-skill-scanner-linux-amd64
Description
使用内置 Rust 引擎审计待安装的 skill 包或归档，并可选桥接外部 scanner。
README (SKILL.md)
Skill Scan Wrapper
当你要在安装一个本地 skill、归档或 release bundle 前做一次快速安全检查时，使用这个 skill。
⚠️ Security Notice
This tool operates locally and requires user trust in the binary you run. Always verify the checksum after downloading. For maximum security, build from source (recommended).
Binary Included
| Property | Value |
|---|---|
| Location | assets/bin/skillscan |
| Version | v0.8.0 |
| Platform | Linux x64 |
| SHA-256 | 864f9a0189268139878c06bce7a127687f9e491a070d7c7345d22932c899bcd8 |
Verify locally before running:
sha256sum assets/bin/skillscan
# Compare output with the SHA-256 value above
This bundled package includes a pre-compiled binary. You can still build from source if you prefer:
git clone https://gitee.com/random_player/cmic-skill-scanner.git
cd cmic-skill-scanner && cargo build --release
前置条件
- 默认不需要任何外部依赖
--upload-url和--engine external功能默认禁用，仅在用户显式配置时启用
信任模型
This is an open-source (MIT-0) package. The binary (bundled or downloaded) is a convenience only — it does not grant any additional trust.
Your options:
| Approach | Trust Requirement | Verification |
|---|---|---|
| Build from source | None (you control everything) | Manual code review |
| Bundled/downloaded binary | You trust the release host | SHA-256 checksum |
What the tool does NOT do by default:
- Does NOT upload data anywhere
- Does NOT connect to the network
- Does NOT access credentials, SSH configs, or environment variables
- Does NOT execute external tools unless you explicitly configure
--engine external
工作流程
- 调用 skillscan：
skillscan review /path/to/target --format markdown
skillscan review /path/to/skills --output-dir /tmp/skillscan-out
- 阅读输出中的：输入类型、完整度、engine 执行状态、findings
网络上传功能 (默认禁用)
⚠️ This feature is completely optional and disabled by default. It requires explicit user configuration via --upload-url.
What gets sent (only when you configure --upload-url):
- A structured JSON report containing detection findings
- An instance identifier you supply via
--instance-id - No skill source code, credentials, or system configuration is ever transmitted
外部引擎集成 (默认禁用)
⚠️ This feature is completely optional and disabled by default. It requires explicit user configuration via --engine external.
Delegates pattern-matching to a user-configured local tool. This runs locally — no remote calls are made.
Permissions Required
| Scope | Reason |
|---|---|
| Read files in target path | To analyze skill source code for patterns |
Write to --output-dir |
To save scan reports locally |
| Execute binary | To run the scanner engine |
| Network (optional) | Only if --upload-url is explicitly configured |
Usage Guidance
Do not run or execute any binary until you confirm where it comes from. Steps to take before installing or using this skill:
- Verify the package actually includes assets/bin/skillscan; if it does not, ask the publisher where to obtain the binary and why it was omitted.
- If you must obtain a binary from the internet, get it only from a tracked release on a trusted host and verify the provided SHA-256 matches the file you downloaded.
- Prefer building from source: clone the referenced repo, inspect the code, and build locally; confirm the built binary's checksum matches the one documented.
- Be cautious enabling --upload-url; confirm (by reading code or by testing in an isolated environment) exactly what fields are transmitted in the JSON report and that no source code or secrets are included.
- Consider running the scanner in an isolated environment (VM/container) until you confirm the origin and contents of the binary.
Because the package claims a bundled executable but does not include it, treat the discrepancy as a red flag and request clarification from the publisher before trusting or executing any binary.
Capability Analysis
Type: OpenClaw Skill
Name: cmic-skill-scanner-linux-amd64
Version: 0.8.0
The skill bundle includes and executes a pre-compiled binary (assets/bin/skillscan) whose source code is not provided within the package, though a Gitee repository is referenced. It also contains built-in functionality for data transmission to remote endpoints via an optional '--upload-url' flag (e.g., scanner.example.com). While the documentation in SKILL.md and INSTALL.md is transparent about these features and their risks, the execution of opaque binaries with network capabilities in an agent environment warrants a suspicious classification.
Capability Tags
Capability Assessment
Purpose & Capability
The SKILL.md and INSTALL.md repeatedly describe a bundled precompiled binary at assets/bin/skillscan and local auditing behavior; however the registry metadata flagged this as an instruction-only skill and the provided file manifest does not include the binary itself. Claiming a bundled executable while not shipping it is an incoherence: users would need to obtain a binary from an external source or build from the referenced repo, which is not made explicit in the package files.
Instruction Scope
Instructions are focused on scanning local skill directories and explicitly state network/upload features are disabled by default. They ask the tool to read target file paths and write output to an output-dir (expected for a scanner). Minor concern: the docs permit optional --upload-url and require an instance-id value, which would transmit structured findings (no source code per doc) — users should verify what exactly would be included before enabling uploads.
Install Mechanism
There is no install spec (lowest technical risk) but SKILL.md claims a bundled binary and provides a SHA-256; the package as presented does not contain the binary path it claims. The SKILL.md suggests cloning a Gitee repo and building from source — pulling code from an external repo is a reasonable option but differs from the 'bundled binary' claim and increases the user's burden to verify the source and checksum. This mismatch raises the chance a user will fetch an external binary from an untrusted location.
Credentials
The skill declares no required environment variables, no primary credential, and no config paths. The documented runtime permissions (read target files, write output-dir, execute binary, optional network only if user supplies --upload-url) are proportional to a local scanning utility.
Persistence & Privilege
The skill is not always-enabled and is user-invocable; it does not request persistent presence or modification of other skills. Autonomous invocation via the model is allowed (platform default) but is not combined with broad or unexpected privileges here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install cmic-skill-scanner-linux-amd64 - After installation, invoke the skill by name or use
/cmic-skill-scanner-linux-amd64 - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.8.0
- Rebranded from "cmic-skill-scanner" to "skillscan-wrapper" with updated descriptions and metadata.
- Documentation overhaul: clearer instructions, detailed trust and security warnings, multilingual content.
- Emphasized that upload/reporting and external engine integrations are disabled by default and require explicit user configuration.
- Updated references to binary version (`v0.8.0`), with consistent SHA-256 and security verification steps.
- Clarified tool permissions, trust model, and provided more explicit build-from-source guidance.
v0.6.4
- Updated to version 0.6.4 with corresponding documentation changes.
- Clarified that the scanner analyzes only skill source code and does not access your system files or credentials.
- Updated binary version references and SHA-256 in documentation.
- Improved description of detected suspicious patterns within skills.
- Removed local-file-read permission requirement from metadata.
v0.6.3
- Updated version to 0.6.3 with revised documentation for improved clarity.
- Repository and author links now point to https://gitee.com/random_player/cmic-skill-scanner.
- Added a "permissions" section, listing required and non-required permissions.
- Reformatted binary information, usage instructions, and malware detection details for better readability.
- Updated links to releases, source, and checksum files.
v0.6.2
- Updated to version 0.6.2 with revised descriptions for clarity.
- Added link to public source code repository and clarified open source status.
- Improved documentation: specified included binary version, platform, and hash, and added instructions to verify the binary.
- Expanded list of security checks and streamlined usage examples.
- Updated metadata with repository URL and reduced tag list.
v0.6.1
- Updated to version 0.6.1 with revised SKILL.md content and metadata.
- Added detailed metadata including author, tags, and triggers for easier discovery and activation.
- Improved skill documentation for clarity, including new usage examples and output descriptions.
- License updated from MIT to MIT-0.
- Path and version references updated for the included binary.
v0.6.0
- Renamed skill to "cmic-skill-scanner" and updated skill description for improved clarity.
- Updated embedded binary version to v0.6.0.
- Documentation now emphasizes security scanning before skill installation and details threats (malware, credential theft, suspicious patterns).
- No external downloader required; package is fully self-contained.
- Minor language and clarity improvements throughout documentation.
v0.5.0
- Updated engine binary to version v0.5.0, with SHA-256 checksum preserved.
- SKILL.md significantly rewritten and simplified; now primarily in Chinese and focused on usage guidance and workflows.
- Clarified usage of optional external scanner (`--engine external/auto`) and details on input requirements.
- Instructions for enterprise integration improved: upload/reporting process and expected outputs are clearly described.
- Old metadata, feature lists, and detailed security guarantee sections removed for brevity.
v0.2.0
- Introduces cmic-skill-scanner version 0.2.0 for Linux (x64), a defensive security tool to scan AI agent skill packages for malware, credential theft, and suspicious patterns before installation.
- Includes a native Rust binary scanner (no external downloads required) with SHA-256 checksum for integrity verification.
- Clearly documents strict security boundaries: does not access personal credentials or send data unless explicitly commanded by the user; no elevated permissions required.
- Adds optional enterprise reporting and support for external scanner engines, both user-controlled via command-line options.
- Provides detailed usage instructions, batch and single skill scanning, report options, and a summary of 31 built-in detection rules.
- Source code and build transparency available for independent verification; alternative platforms (macOS ARM, Linux ARM) referenced.
Metadata
Frequently Asked Questions
What is CMIC Skill Scanner (Linux x64)?
使用内置 Rust 引擎审计待安装的 skill 包或归档，并可选桥接外部 scanner。 It is an AI Agent Skill for Claude Code / OpenClaw, with 146 downloads so far.
How do I install CMIC Skill Scanner (Linux x64)?
Run "/install cmic-skill-scanner-linux-amd64" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is CMIC Skill Scanner (Linux x64) free?
Yes, CMIC Skill Scanner (Linux x64) is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does CMIC Skill Scanner (Linux x64) support?
CMIC Skill Scanner (Linux x64) is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created CMIC Skill Scanner (Linux x64)?
It is built and maintained by cyzlmh (@cyzlmh); the current version is v0.8.0.
More Skills