Shadows MCP Forge
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This instruction-only skill is coherently focused on helping users build MCP servers, with expected but user-noticeable package installation, credential, and persistent MCP configuration implications.
This skill appears safe to use as an instruction-only MCP server builder. Before installing generated servers, review the code, dependencies, tools exposed to the agent, and any API keys or service permissions you configure.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Installing these dependencies adds third-party code to the user's development environment.
The skill instructs users to install third-party packages from public registries. This is expected for an MCP server builder, and the artifact explicitly tells users to review package names.
Commands suggested: `pip install fastmcp`, `npm install @modelcontextprotocol/sdk` — these install packages from public registries. Review package names before running.
Review package names and sources before installing, and prefer pinned versions in real projects.
Generated MCP servers may use service credentials, which could let an AI agent access or modify third-party data depending on the token's scope.
The sample MCP configuration shows passing an API key to generated servers. This is purpose-aligned for external integrations and is paired with guidance not to hardcode secrets.
"env": { "API_KEY": "from-env-or-secrets" }Use narrowly scoped API keys, store them in environment variables or a secret manager, and avoid giving generated tools broader permissions than needed.
Running generated server code can execute local project code and connect it to an AI client.
The skill provides commands for running and installing generated MCP server code. This is central to the stated purpose and is presented as user-directed usage, not hidden automatic execution.
fastmcp dev server.py # Install in Claude/OpenClaw fastmcp install server.py --name "My Service"
Review generated code before running or installing it, especially if it accesses files, credentials, databases, or external services.
A generated MCP server may remain available to the AI client after setup, continuing to expose whatever tools and credentials it was configured with.
The configuration example shows adding a persistent MCP server entry to an AI client. This is disclosed and expected for the skill, but it persists beyond a single chat unless removed.
For OpenClaw/Claude Desktop: { "mcpServers": { "my-service": { "command": "python", "args": ["-m", "my_mcp_server.server"] } } }Install only MCP servers you trust, document what tools they expose, and remove or disable them when no longer needed.