← Back to search

Search • Sprawl Radar • TrustOps • Gateway • Ledger • Mandates • Hosting • Agent Commerce • Docs Compiler • Demo • Trust Index • Discovery Sources • Submit Repo • GitHub Action • Plans • Intelligence API • Verified Profiles • Shortlists • Rankings • Compare • Badges • Claim • ChatGPT compatible • Claude compatible • Healthcare • Digests • RSS • RSS XML • Backfill • MCP API

MCP VERIFY MCP API

Use Verify as an MCP server

Verify exposes an agent-native MCP surface for discovery, trust reports, policy export, permission decisions, mandates, ledger evidence, hosting controls, and docs compilation.

Endpoint

https://verify.sentinelsignal.io/mcp

JSON-RPC over HTTP. The human-readable page is GET /mcp; the MCP transport is POST /mcp.

Server card

/.well-known/mcp/server-card.json

Public metadata for registries and clients.

Agent docs

llms.txt

Discovery index with tool list, canonical category URLs, API docs, and example prompts.

Registry presence

Smithery

Official registry metadata is tracked in-repo as mcp-registry/verify/server.json.

Available tools

ChatGPT-compatible alias returning id, title, and url.

fetch

ChatGPT-compatible alias returning a full item with id, title, text, url, and metadata.

search_servers

Rank candidates by capability, auth preference, client target, and risk tolerance.

recommend_servers

Turn a plain-language task into top MCP matches with install config.

get_server_report

Return the full machine-readable verify report for one server.

compare_servers

Compare up to four servers across score, auth, freshness, tools, and verdict.

route_task

Choose the best MCP server/tool path for a task. This is selection, not final permission.

decide_agent_call

Allow, deny, or require approval for a specific attempted call by a specific agent and user; writes Ledger evidence.

create_mandate

Create signed delegated authority with scope, budget, validity window, and tool constraints.

get_evidence_pack

Export Ledger invocations, hash-chained audit events, decisions, outcomes, mandates, and costs.

compile_docs

Generate MCP drafts, server-card.json, llms.txt, tool schemas, examples, policy metadata, and hosted docs pages.

export_policy

Export a JSON policy for one MCP server with freshness, scopes, and tool decisions.

get_subscription_options

Return alert types, channels, watch scopes, and subscription endpoints.

get_gateway_options

Return Gateway semantics, /v1/route versus /v1/decide, mandate inputs, Ledger hooks, and decision fields.

get_hosting_options

Return Verify Hosted MCP runtime and sandbox hosting capabilities.

list_agents

List Agent Sprawl Radar inventory with owners, risk levels, tools, and data classes.

get_agent

Fetch one discovered agent with findings, tools, data access, and drift.

score_agent

Score a normalized agent payload or existing registry agent.

explain_agent_risk

Explain findings and deterministic recommended action.

list_high_risk_agents

List high and critical risk agents for review.

get_drift_report

Return latest drift events across agent snapshots.

generate_remediation_plan

Produce deterministic remediation steps for risky or drifting agents.

Governance tools

Gateway

decide_agent_call

Permission decision point for one attempted call by a known agent, user, server, tool, payload, and mandate.

Gateway metadata

get_gateway_options

Discover decision inputs, outputs, endpoint semantics, and how Gateway connects to Mandates and Ledger.

Mandates

create_mandate

Create signed delegated authority records with scope, budget, validity, and tool constraints.

Ledger

get_evidence_pack

Export replayable evidence for decisions, mandates, policy snapshots, outcomes, and costs.

Docs Compiler

compile_docs

Generate agent-readable MCP surfaces and policy metadata from docs.

Agent Sprawl Radar tools

These tools are live on POST /mcp and listed in the public server card for assistants that need to inspect internal agent inventory, risk, drift, and remediation evidence.

Inventory

list_agents

List discovered agents, MCP servers, bots, workflows, owners, risk levels, tool counts, and data classes.

Detail

get_agent

Return one discovered agent with tools, data access, findings, latest drift, and recommended action.

Risk

score_agent explain_agent_risk

Score a normalized payload or explain why an existing registry agent has its current risk level.

Drift

get_drift_report

Return material drift events across latest agent snapshots.

Review

list_high_risk_agents generate_remediation_plan

Find high/critical agents and produce deterministic remediation steps.

Route versus decide

/v1/route

Choose the best server/tool

Use this before execution when an agent has a task and needs a recommended MCP path.

/v1/decide

Authorize one attempted call

Use this at execution time when the agent, user, server, tool, payload, cost, and mandate are known.

Quick start

Claude Desktop / generic wrapper

{
  "mcpServers": {
    "verify": {
      "command": "npx",
      "args": ["mcp-remote", "https://verify.sentinelsignal.io/mcp"]
    }
  }
}

Search example

curl -sS https://verify.sentinelsignal.io/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_servers","arguments":{"capabilities":["healthcare"],"auth_preference":"oauth","client_target":"openai_connectors","risk_tolerance":"high","limit":3}}}'

Gateway example

curl -sS https://verify.sentinelsignal.io/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"decide_agent_call","arguments":{"agent_id":"agent_123","user_id":"user_456","server_id":"io.sentinelsignal/scoring","tool":"score_workflow","payload_classification":["phi"],"requested_cost":0.02}}}'