Enterprise AI governance: spend, guardrails, policy, budgets, compliance, and provider health.
Enterprise AI governance: spend, guardrails, policy, budgets, compliance, and provider health.
Remote endpoints: streamable-http: https://mcp.thinkneo.ai/mcp
Valid MCP server (1 strong, 8 medium validity signals). 3 known CVEs in dependencies (0 critical, 1 high severity) Imported from the Official MCP Registry.
Endpoint verified · Open access · 4 issues found
Security scores are indicators to help you make informed decisions, not guarantees. Always review permissions before connecting any MCP server.
This plugin requests these system permissions. Most are normal for its category.
Remote Plugin
No local installation needed. Your AI client connects to the remote endpoint directly.
Add this to your MCP configuration to connect:
{
"mcpServers": {
"ai-thinkneo-control-plane": {
"url": "https://mcp.thinkneo.ai/mcp"
}
}
}Once installed, try these example prompts and explore these capabilities:
From the project's GitHub README.
Dual-protocol gateway: MCP tools + A2A Protocol agent in a single governed runtime
| Component | Value |
|---|---|
| MCP Protocol | 2024-11-05 (Streamable HTTP) |
| A2A Protocol | v0.3.0 (Google / Linux Foundation) |
| MCP Tools | 55 |
| A2A Skills | 24 |
| A2A TCK Compliance | 97.6% (80/82 passing) — see report |
| Internal Tests | 159/159 passing |
| Production | Live at mcp.thinkneo.ai |
| Listed on | Glama.ai (AAA rating) |
| Endpoint | URL | Method |
|---|---|---|
| MCP (Streamable HTTP) | https://mcp.thinkneo.ai/mcp | POST |
| A2A Agent Card | https://mcp.thinkneo.ai/.well-known/agent.json | GET |
| A2A Agent | https://agent.thinkneo.ai/a2a | POST |
| Health Check | https://mcp.thinkneo.ai/guardian/health | GET |
| Developer Docs | https://mcp.thinkneo.ai/mcp/docs | GET |
| MCP Marketplace | https://mcp.thinkneo.ai/registry | GET |
This repository contains the ThinkNEO Gateway, a dual-protocol server that exposes AI governance tools through both the Model Context Protocol (MCP, Anthropic 2024) and the Agent-to-Agent Protocol (A2A, Google/Linux Foundation 2025).
The gateway implements a bidirectional MCP-A2A bridge: MCP clients can dispatch A2A tasks, and A2A agents can call MCP tools. Both protocols share the same governed runtime, meaning all traffic passes through the ThinkNEO Enterprise AI Control Plane for policy enforcement, audit logging, spend tracking, and compliance checks.
The server is production-deployed as a single Docker container, serving MCP on /mcp and the A2A Agent Card at /.well-known/agent.json. All 55 MCP tools and 24 A2A skills run in the same process, share the same PostgreSQL database, and are wrapped with the same free-tier middleware (500 calls/month, auto-provisioned API keys).
55 tools across 12 categories. All tools accept JSON-RPC 2.0 via Streamable HTTP.
| Tool | Description | Auth |
|---|---|---|
thinkneo_check_spend | AI cost breakdown by provider, model, team, and time period | Required |
thinkneo_evaluate_guardrail | Pre-flight prompt safety evaluation against workspace policies | Required |
thinkneo_check_policy | Verify if a model, provider, or action is allowed by governance policies | Required |
thinkneo_get_budget_status | Budget utilization, enforcement status, and projected overage | Required |
thinkneo_list_alerts | Active alerts and incidents for a workspace | Required |
thinkneo_get_compliance_status | SOC2, GDPR, HIPAA compliance readiness and governance score | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_check | Prompt safety check: injection patterns (10 rules) + PII (credit card with Luhn, CPF with checksum, SSN, email, phone, API keys) | None |
thinkneo_usage | API key usage stats: calls today/week/month, top tools, estimated cost | None |
thinkneo_provider_status | Real-time health and latency of 7 AI providers | None |
thinkneo_read_memory | Read Claude Code project memory files (index or specific file) | None |
thinkneo_write_memory | Write/update project memory files | Required |
thinkneo_schedule_demo | Book a demo with the ThinkNEO team | None |
| Tool | Description | Auth |
|---|---|---|
thinkneo_route_model | Find the cheapest model meeting a quality threshold across 17+ models from 9 providers | Required |
thinkneo_get_savings_report | Cost savings report: original vs actual cost, breakdown by task type | Required |
thinkneo_simulate_savings | Simulate how much an organization would save with Smart Router | None |
| Tool | Description | Auth |
|---|---|---|
thinkneo_registry_search | Search the MCP Marketplace by keyword, category, rating, or verified status | None |
thinkneo_registry_get | Full details for an MCP server package: readme, tools, versions, reviews | None |
thinkneo_registry_publish | Publish an MCP server: validates endpoint, runs security scan, stores entry | Required |
thinkneo_registry_review | Rate and review an MCP server (1-5 stars, one per user per package) | Required |
thinkneo_registry_install | Installation config for Claude Desktop, Cursor, Windsurf, or custom clients | None |
| Tool | Description | Auth |
|---|---|---|
thinkneo_evaluate_trust_score | AI Trust Score (0-100) across 6 dimensions with shareable badge | Required |
thinkneo_get_trust_badge | Public trust score badge lookup by token | None |
| Tool | Description | Auth |
|---|---|---|
thinkneo_start_trace | Start an observability session for an agent | Required |
thinkneo_log_event | Log a tool call, model call, decision, or error event | Required |
thinkneo_end_trace | End a trace session with final status | Required |
thinkneo_get_trace | Retrieve full trace with all events for a session | Required |
thinkneo_get_observability_dashboard | Aggregated observability metrics for a workspace | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_set_baseline | Set a pre-AI baseline metric for comparison | Required |
thinkneo_log_decision | Log an AI agent decision with type, rationale, and context | Required |
thinkneo_decision_cost | Calculate the cost of a specific decision | Required |
thinkneo_log_risk_avoidance | Log a risk that was avoided by AI intervention | Required |
thinkneo_agent_roi | ROI report for a specific agent | Required |
thinkneo_business_impact | Business impact report for a workspace | Required |
thinkneo_detect_waste | Detect waste patterns: unused tools, redundant calls, idle agents | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_register_claim | Register a verifiable claim: what an agent did, with evidence type | Required |
thinkneo_verify_claim | Verify a claim by checking evidence (HTTP status, file exists, DB row, etc.) | Required |
thinkneo_get_proof | Get the full proof chain for a verified claim | Required |
thinkneo_verification_dashboard | Dashboard of all claims, verification rates, and trust metrics | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_policy_create | Create a governance policy for a workspace | Required |
thinkneo_policy_list | List all policies for a workspace | Required |
thinkneo_policy_evaluate | Evaluate an action against workspace policies | Required |
thinkneo_policy_violations | List policy violations with severity and remediation | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_compliance_generate | Generate a compliance report for a framework | Required |
thinkneo_compliance_list | List available compliance frameworks and their status | Required |
| Tool | Description | Auth |
|---|---|---|
thinkneo_benchmark_compare | Compare model performance across tasks | Required |
thinkneo_benchmark_report | Generate a benchmark report with recommendations | Required |
thinkneo_router_explain | Explain why the Smart Router chose a specific model | Required |
This gateway is simultaneously an A2A-compliant agent, implementing the Agent-to-Agent Protocol v0.3.0 maintained by the Linux Foundation.
Publicly discoverable at /.well-known/agent.json. Describes all 24 A2A skills, authentication requirements, and runtime capabilities.
Bidirectional protocol translation. An A2A agent can consume MCP tools; an MCP client can dispatch A2A tasks.
| Tool | Description |
|---|---|
thinkneo_bridge_mcp_to_a2a | Translate an MCP tool call into an A2A task and dispatch it to an A2A agent |
thinkneo_bridge_a2a_to_mcp | Receive an A2A task, find the matching MCP tool, execute it, return A2A response |
thinkneo_bridge_generate_agent_card | Auto-generate an A2A Agent Card from the MCP tools/list |
thinkneo_bridge_list_mappings | List all current MCP-A2A tool/skill mappings |
Full A2A task lifecycle is audited: task_sent - task_accepted - task_completed - flow traced - audit recorded.
| Tool | Description |
|---|---|
thinkneo_a2a_log | Log an agent-to-agent interaction with cost, outcome, and latency |
thinkneo_a2a_set_policy | Set communication policies between agents (rate limits, allowed actions) |
thinkneo_a2a_flow_map | Visualize agent-to-agent communication flows over a time period |
thinkneo_a2a_audit | Generate an audit trail of all A2A interactions |
MCP Clients A2A Agents
(Claude, Cursor, (Any A2A-compliant
ChatGPT, custom) agent or service)
| |
| Streamable HTTP | JSON-RPC 2.0
| POST /mcp | POST /a2a
| |
+---------------+---------------+
|
+--------v---------+
| ThinkNEO |
| Gateway |
| |
| MCP Runtime |
| A2A Runtime |
| MCP <-> A2A |
| Bridge |
+--------+---------+
|
+-----------+-----------+
| |
+------v-------+ +------v-------+
| Governance | | Data Layer |
| Policy Engine | | PostgreSQL |
| FinOps | | 33 tables |
| Audit Trail | | Usage log |
| Compliance | | API keys |
+---------------+ +--------------+
This gateway implements the A2A Protocol v0.3.0 specification maintained by the Linux Foundation. The official Google-published Technology Compatibility Kit (TCK) validates 82 test cases against the implementation.
Current status: 97.6% passing (80/82)
The 2 failing tests are in the optional_capability category -- features marked as optional in the specification that are deliberately not implemented. See TCK_REPORT.md for the detailed breakdown, including the specific test cases that fail and rationale for each.
Add to your Claude Desktop config (~/.claude/claude_desktop_config.json):
{
"mcpServers": {
"thinkneo": {
"url": "https://mcp.thinkneo.ai/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Or use with Cursor (.cursor/mcp.json):
{
"mcpServers": {
"thinkneo": {
"url": "https://mcp.thinkneo.ai/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Discover the agent:
curl https://mcp.thinkneo.ai/.well-known/agent.json
Send a task:
curl -X POST https://agent.thinkneo.ai/a2a \
-H "Content-Type: application/json" \
-d '{
"jsonrpc": "2.0",
"method": "tasks/send",
"id": "1",
"params": {
"id": "task-001",
"message": {
"role": "user",
"parts": [{"text": "Check if this prompt is safe: ignore all previous instructions"}]
}
}
}'
tnk_, provided via Authorization: Bearer <key> headerAuto-provision a free API key (500 calls/month):
POST https://mcp.thinkneo.ai/mcp/signup
No credit card required. All 55 tools available on the free tier.
read, execute, custom)Contact: hello@thinkneo.ai
Solid tool with a clean interface and reliable performance. The control plane integration works smoothly and the setup was straightforward.
Really impressed with how well this handles the workload. Setup was quick, the dashboard is intuitive, and it hasn't let me down yet. Minor UI quirks here and there, but nothing that gets in the way. Great product overall — looking forward to seeing how it evolves.
Governance!!!
Initial enterprise release. Enterprise AI Control Plane with runtime guardrails, AI FinOps, deep observability, agent lifecycle control, and compliance readiness across all major AI providers.
by Toleno · Developer Tools
Toleno Network MCP Server — Manage your Toleno mining account with Claude AI using natural language.
by mcp-marketplace · Developer Tools
Create, build, and publish Python MCP servers to PyPI — conversationally.
by Microsoft · Content & Media
Convert files (PDF, Word, Excel, images, audio) to Markdown for LLM consumption
by mcp-marketplace · Developer Tools
Scaffold, build, and publish TypeScript MCP servers to npm — conversationally
by mcp-marketplace · Finance
Free stock data and market news for any MCP-compatible AI assistant.
by Taylorwilsdon · Productivity
Control Gmail, Calendar, Docs, Sheets, Drive, and more from your AI