← Back to Skills Marketplace
Switchyard Runtime Diagnostics
by
Yifeng[Terry] Yu
· GitHub ↗
· v1.0.0
· MIT-0
73
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install switchyard-runtime-diagnostics
Description
Guide to install, connect, and use Switchyard's read-only MCP runtime diagnostics for analyzing provider or runtime boundaries safely.
Usage Guidance
This packet is coherent for local, read-only Switchyard diagnostics, but it requires cloning and running a third-party GitHub repo with pnpm. Before installing or following its instructions: 1) review the referenced repository (package.json scripts, pnpm install effects, any postinstall hooks) to ensure you trust it; 2) run the demo in an isolated environment (container or VM) to limit blast radius; 3) replace /ABSOLUTE/PATH/TO/SWITCHYARD carefully and avoid running commands as root; 4) if you cannot review the repo, do not run pnpm install/run on sensitive hosts; 5) the skill claims read-only behavior — verify the MCP endpoints and tests in a safe environment to confirm they only expose runtime diagnostics and do not perform writes or exfiltrate data.
Capability Analysis
Type: OpenClaw Skill
Name: switchyard-runtime-diagnostics
Version: 1.0.0
The skill bundle instructs the agent to clone an external repository (https://github.com/xiaojiou176-open/Switchyard.git) and execute 'pnpm install' followed by 'pnpm run' to initialize an MCP server (references/INSTALL.md, references/OPENCLAW_MCP_CONFIG.json). While these actions are consistent with the stated goal of setting up runtime diagnostics, they involve high-risk operations including remote code execution and shell access. No explicit evidence of malicious intent or data exfiltration was found, but the requirement for the agent to download and run arbitrary third-party scripts warrants a suspicious classification.
Capability Assessment
Purpose & Capability
Name/description match the contents: the skill is a read-only diagnostics packet that teaches an agent how to attach a local MCP server and run specific read-only queries. Nothing in the files asks for unrelated credentials, system-wide config, or capabilities beyond starting and querying a local MCP surface.
Instruction Scope
SKILL.md and references direct the agent to clone a GitHub repo and run pnpm scripts to launch a local MCP server, then call read-only MCP tools. This stays within the stated diagnostics scope, but it does instruct executing code from an external repository and running local commands (pnpm install/run), which can have side effects if the repository or its scripts are malicious or untrusted.
Install Mechanism
No install spec in the skill bundle itself (instruction-only). The instructions rely on cloning a third-party GitHub repo and running pnpm, which is a common but higher-risk install approach because it pulls and executes upstream code. The repo URL is explicit (github.com/xiaojiou176-open/Switchyard.git) — not a shortener or IP, but you should verify the repo before cloning and running.
Credentials
The skill declares no required environment variables, no credentials, and no config paths beyond repo-local MCP config snippets. The requested access is proportional to the stated task: launching and querying a local read-only runtime. There are no unrelated secrets requested.
Persistence & Privilege
always is false and the packet is instruction-only; it does not request persistent presence or modification of other skills. Autonomous invocation is allowed by default (platform behavior) but not unusual here and not combined with other privilege escalation indicators.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install switchyard-runtime-diagnostics - After installation, invoke the skill by name or use
/switchyard-runtime-diagnostics - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of switchyard-runtime-diagnostics skill:
- Provides installation and usage instructions for Switchyard's read-only MCP runtime diagnostics.
- Details safe tools for examining runtime and catalog state without making changes.
- Guides diagnosis of providers and runtime boundaries, with specific guardrails for safety.
- Includes step-by-step references for configuration and troubleshooting.
- Emphasizes read-only, diagnostic use to ensure claims stay partial and non-disruptive.
Metadata
Frequently Asked Questions
What is Switchyard Runtime Diagnostics?
Guide to install, connect, and use Switchyard's read-only MCP runtime diagnostics for analyzing provider or runtime boundaries safely. It is an AI Agent Skill for Claude Code / OpenClaw, with 73 downloads so far.
How do I install Switchyard Runtime Diagnostics?
Run "/install switchyard-runtime-diagnostics" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Switchyard Runtime Diagnostics free?
Yes, Switchyard Runtime Diagnostics is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Switchyard Runtime Diagnostics support?
Switchyard Runtime Diagnostics is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Switchyard Runtime Diagnostics?
It is built and maintained by Yifeng[Terry] Yu (@xiaojiou176); the current version is v1.0.0.
More Skills