check_token
Request the current token-risk decision and supporting evidence.
Open source · MCP
Give an MCP-compatible agent a pre-trade token-risk tool.
VerdictSwarm MCP is an MIT-licensed Python wrapper that gives compatible AI clients three current tools: check_token, exit_sim, and get_pricing.
Release 0.2.1 · Published August 3, 2026.
Tool surface
Three tools, one external risk service
exit_sim
Estimate intended-size exit behavior where the underlying capability is supported.
get_pricing
Read current tool and payment information from the service.
The MCP package is a stateless wrapper. It sends requests to the external VerdictSwarm API; it does not run token intelligence locally. Integrators should follow the current API documentation for support boundaries and decision semantics.
Run it yourself
PyPI or uvx
Choose the package when you want the runtime in your own environment and control its version, secrets, process lifecycle, and client configuration.
pip install verdictswarm-mcpHosted discovery
Smithery
Choose Smithery when your client supports its hosted connection flow and you prefer a managed discovery and connection surface over running a local wrapper process.
Open the Smithery server →Source and trust
Inspect the wrapper before installing it
The source repository includes client examples, tests, a contribution guide, and a private security-reporting path. Keep API keys in the client's secret configuration; never paste them into prompts or commit them to a repository.
Not sure whether MCP is the right path?
Compare the direct API, a self-run MCP server, Smithery, and the OpenClaw-native skill before deciding who should own the runtime and credentials.
Compare integration paths →