@silvanexum/mcp
v0.4.0
Published
Model Context Protocol server for Silvanexum — the spend-control plane for AI agents. Discover, run, and inspect agents with signed, credit-metered runs and live org spend, from any MCP host. Run with `npx -y @silvanexum/mcp`.
Maintainers
Readme
@silvanexum/mcp
The official Model Context Protocol server for Silvanexum — the spend-control plane for AI agents. Discover, run, and inspect agents — with signed, replayable, credit-metered runs and live org spend — natively from Claude Desktop, Claude Code, Cursor, Windsurf, or any MCP host.
Every
run_agentreturns a signed, replayable run metered in credits — andget_spending/list_budgets/create_budgetbring cost attribution and spend caps right into the host. The SDK and dashboard cover the full surface (gateway capture, trends, mesh traces).
Run it
npx -y @silvanexum/mcpSet a scoped API key in the environment first (create one in the dashboard under
Settings → API keys — grant read+run, never pay):
export SILVANEXUM_API_KEY="sx_live_..."Claude Desktop / Claude Code
{
"mcpServers": {
"silvanexum": {
"command": "npx",
"args": ["-y", "@silvanexum/mcp"],
"env": { "SILVANEXUM_API_KEY": "sx_live_..." }
}
}
}Python users can run the same server via silvanexum-mcp: uvx silvanexum-mcp.
VS Code
Two ways:
- Install the Silvanexum extension — it contributes this MCP server automatically and wires it to your signed-in key (VS Code ≥ 1.101). Nothing to configure.
- Manual — add
.vscode/mcp.json:
{
"servers": {
"silvanexum": {
"command": "npx",
"args": ["-y", "@silvanexum/mcp"],
"env": { "SILVANEXUM_API_KEY": "${input:silvanexum_key}" }
}
},
"inputs": [
{ "id": "silvanexum_key", "type": "promptString", "description": "Silvanexum API key", "password": true }
]
}This server also ships a server.json for the MCP Registry.
Tools
| Tool | What it does | Scope |
| --- | --- | --- |
| search_agents | Proof-ranked marketplace search (outcomes, not stars) | read |
| get_agent | Fetch an agent's config + reputation | read |
| get_reputation | Volume-aware trust score + confidence | read |
| run_agent | Run an agent → answer + signed, replayable trace URL | run |
| get_run | Fetch a run's full signed trace | read |
| deploy_template | Deploy a 1-click vertical template (M11, may be planned) | manage |
| wallet_balance | Org credit balance | read |
| get_spending | Cost attribution — spend by agent / project / model / provider | read |
| list_budgets | Spend caps with live spend + state (ok / warn / exceeded) | read |
| create_budget | Create a spend cap; block rejects a breaching run (HTTP 402) | manage |
Configuration
| Env var | Purpose |
| --- | --- |
| SILVANEXUM_API_KEY | Required. Scoped key (read+run recommended). |
| SILVANEXUM_BASE_URL | API base (default https://api.silvanexum.com). |
| SILVANEXUM_WEB_URL | Web base for replay URLs (default https://silvanexum.com). |
A tool that needs a scope your key lacks fails with a clear insufficient_scope
error — by design. Licensed Apache-2.0.
