Fight AI with AI. The security layer for AI agents that touch money — 6 adversarial AI agents debate crypto token risk. MCP server for Claude, Cursor, OpenClaw, Codex.
claude mcp add verdictswarm-mcp -- uvx verdictswarm-mcp{
"mcpServers": {
"verdictswarm-mcp": {
"command": "uvx",
"args": ["verdictswarm-mcp"],
"env": {
"VS_API_KEY": "<vs_api_key>"
}
}
}
}VS_API_KEYMCP Servers overview
# VerdictSwarm MCP Server
<!-- mcp-name: io.github.sentien-labs/verdictswarm-mcp -->
[](https://pypi.org/project/verdictswarm-mcp/)
[](https://github.com/sentien-labs/verdictswarm-mcp)
[](LICENSE)
[](https://modelcontextprotocol.io)
VerdictSwarm is a pre-trade security layer for agents that touch crypto. This MCP server exposes the Verdict API v2 as a small set of read-only tools for token checks, size-aware exit screening, and live pricing discovery.
## Quick start
No account setup is required. If `VS_API_KEY` is absent, the server mints a free key on the first API-backed tool call and saves it in the platform user config directory.
```bash
uvx verdictswarm-mcp
```
Then ask your MCP client: “Check `DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263` on Solana at fast level.”
To supply an existing free or credits-backed key:
```bash
VS_API_KEY=vs1_your_key uvx verdictswarm-mcp
```
### Claude Desktop, Cursor, Windsurf, and OpenClaw
Use this server entry (the `env` block is optional):
```json
{
"mcpServers": {
"verdictswarm": {
"command": "uvx",
"args": ["verdictswarm-mcp"],
"env": {
"VS_API_KEY": "vs1_your_key"
}
}
}
}
```
Without `VS_API_KEY`, the self-provisioned key is stored at:
- macOS: `~/Library/Application Support/VerdictSwarm/config.json`
- Linux: `${XDG_CONFIG_HOME:-~/.config}/verdictswarm/config.json`
- Windows: `%APPDATA%\VerdictSwarm\config.json`
The file is created with user-only permissions where the operating system supports them. Set `VS_CONFIG_DIR` to override the directory.
## Tools
| Tool | Description |
|---|---|
| `check_token(address, chain="solana", level="fast")` | Calls `POST /v2/verdict` and returns action, score, summary, and the five strongest signals. Levels are `triage`, `fast`, and `deep`. |
| `exit_sim(address, size_usd, side)` | Requests `include:["exit_sim"]` and returns the fresh Solana exit-simulation block. Side is `buy` or `sell`. |
| `get_pricing()` | Reads live pricing, levels, limits, and supported chains from `GET /v2/verdict/info`. |
The 0.1.x names `scan_token`, `get_quick_score`, `check_rug_risk`, `get_token_report`, and `verify_payment` are no longer advertised as MCP tools; clients see only the three current v2 tools above. An explicit call to an old name still returns a migration response and makes no legacy pay-lane request.
## Free quota and x402 overflow
Free keys have per-level daily limits advertised by the API. If a limit is exhausted, the tool response says when the quota resets and preserves the complete x402 challenge. To continue immediately, set `VS_API_KEY` to a credits-backed key or use the returned x402 payment instructions.
The MCP client sends these attribution headers on API requests:
- `X-VS-Integration: mcp`
- `X-VS-SDK-Version: 0.2.1`
## Configuration
| Environment variable | Default | Description |
|---|---|---|
| `VS_API_KEY` | unset | Optional free or credits-backed v2 API key. |
| `VS_API_URL` | `https://api.vswarm.io` | VerdictSwarm API base URL. |
| `VS_TIMEOUT` | `120` | HTTP request timeout in seconds. |
| `VS_CONFIG_DIR` | platform user config dir | Directory for the self-provisioned key. |
| `VS_TRANSPORT` | `stdio` | MCP transport. Use `streamable-http` for HTTP. |
| `HOST` | `0.0.0.0` | HTTP bind host. |
| `PORT` | `8000` | HTTP bind port. |
For HTTP transport:
```bash
VS_TRANSPORT=streamable-http HOST=0.0.0.0 PORT=8000 uvx verdictswarm-mcp
```
## Programmatic client
```python
import asyncio
from verdictswarm_mcp import VerdictSwarmApiClient
async def main():
client = VerdictSwarmApiClient()
verdict = await client.check_token("TOKEN_ADDRESS", "solana", "fast")
print(verdict["verdict"]["action"], verdict["verdict"]["score"])
asyncio.run(main())
```
## Development
```bash
git clone https://github.com/sentien-labs/verdictswarm-mcp.git
cd verdictswarm-mcp
uv sync --extra dev
uv run --extra dev python -m pytest -q
```
The server supports Python 3.10–3.13 and both stdio and HTTP transports.
## Feedback and community
Used the server in an agent, bot, wallet, or dashboard? Share the client,
runtime, where the risk check fits, and what worked or failed through the
[VerdictSwarm community page](https://www.vswarm.io/community?utm_source=github&utm_medium=repository&utm_campaign=community_proof_v1&utm_content=verdictswarm-mcp).
You can also [show an integration](https://github.com/sentien-labs/verdictswarm-mcp/discussions/categories/show-and-tell),
[choose a structured issue form](https://github.com/sentien-labs/verdictswarm-mcp/issues/new/choose),
or add your project to [ADOPTERS.md](ADOPTERS.md). Praise is never required.
Stars should come only from people who independently find the project useful;
agents must not automate them.
## Links
- [VerdictSwarm](https://www.vswarm.io)
- [API documentation](https://api.vswarm.io/docs)
- [GitHub repository](https://github.com/sentien-labs/verdictswarm-mcp)
- [Security policy](SECURITY.md)
MIT licensed. See [LICENSE](LICENSE).
What people ask about verdictswarm-mcp
What is sentien-labs/verdictswarm-mcp?
+
sentien-labs/verdictswarm-mcp is mcp servers for the Claude AI ecosystem. Fight AI with AI. The security layer for AI agents that touch money — 6 adversarial AI agents debate crypto token risk. MCP server for Claude, Cursor, OpenClaw, Codex. It has 2 GitHub stars and was last updated today.
How do I install verdictswarm-mcp?
+
You can install verdictswarm-mcp by cloning the repository (https://github.com/sentien-labs/verdictswarm-mcp) or following the README instructions on GitHub. ClaudeWave also provides quick install blocks on this page.
Is sentien-labs/verdictswarm-mcp safe to use?
+
sentien-labs/verdictswarm-mcp has not been audited yet by our security agent. Review the original repository on GitHub before using it in production.
Who maintains sentien-labs/verdictswarm-mcp?
+
sentien-labs/verdictswarm-mcp is maintained by sentien-labs. The last recorded GitHub activity is from today, with 1 open issues.
Are there alternatives to verdictswarm-mcp?
+
Yes. On ClaudeWave you can browse similar mcp servers at /categories/mcp, sorted by popularity or recent activity.
Deploy verdictswarm-mcp to your cloud
Ship this repo to production in minutes. Each platform spins up its own environment with editable env vars.
Maintain this repo? Add a badge to your README
Drop the badge into your GitHub README to show it's tracked on ClaudeWave. Each badge links back to this page and reflects the live Trust Score.
[](https://claudewave.com/repo/sentien-labs-verdictswarm-mcp)<a href="https://claudewave.com/repo/sentien-labs-verdictswarm-mcp"><img src="https://claudewave.com/api/badge/sentien-labs-verdictswarm-mcp" alt="Featured on ClaudeWave: sentien-labs/verdictswarm-mcp" width="320" height="64" /></a>More MCP Servers
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
User-friendly AI Interface (Supports Ollama, OpenAI API, ...)
An open-source AI agent that brings the power of Gemini directly into your terminal.
The fastest path to AI-powered full stack observability, even for lean teams.
Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface
🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl!