A2A Governance Bridge MCP server. Tools: verify agent compliance, authorize a2a transaction, get trust registry. Built by MEOK AI Labs.
Free
Tools
thinkneo_check_spend
Check AI spend summary for a workspace, team, or project. Returns real cost breakdown by provider, model, and time period from the ThinkNEO AI gateway.
ArgumentsA call is made on an account: it counts against an allowance and the publisher sees it, which is why this one asks who you are first.
—
p95 latency
—
call success
—
uptime 90d
0
calls last 7d
What it does
Enterprise AI governance: spend, guardrails, policy, budgets, compliance, and provider health.
Quickstart
# 1 — install (mcprush login holds a key from your dashboard)
npx mcprush@latest add control-plane-mcp
# 2 — ask your agent something
> Enterprise AI governance: spend, guardrails, policy, budgets, compliance, and provider health.
Control Plane is free: there is no plan to choose, no cap to set and nothing that can bill you.
Publishernot claimed
MCAll 1 server
mcp.thinkneo.ai1 server
1 category
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Installation
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. Nothing here runs on your machine — this server runs on the publisher’s own infrastructure behind our gateway, and what you install is the connection to it.
Claude Codedeclared by this listing
This is a public server: you run it yourself and this marketplace is not in the path. Claude Code registers it in one command.
claude mcp add --transport http control-plane-mcp https://mcp.thinkneo.ai/mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
Free to install, and it runs from its publisher’s own package: no account, no key, no card.
What it writes
One config entry pointing at the gateway. The server itself runs on the publisher’s own infrastructure, so nothing from this listing executes on your machine.
What you approve
68 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Check AI spend summary for a workspace, team, or project. Returns real cost breakdown by provider, model, and time period from the ThinkNEO AI gateway.
Takes no parameters.
Example callreturns →
Run a governed LLM completion through the ThinkNEO AI gateway. The request is authorized, classified, and policy-checked against your workspace governance BEFORE any provider is called — a blocked prompt never reaches the model. Tenant/workspace are derived from your API key.
Takes no parameters.
Example callreturns →
Read Claude Code project memory files. Without arguments, returns the MEMORY.md index listing all available memories. With a filename argument, returns the full content of that specific memory file. Use this to access project context, user preferences, feedback, and reference notes persisted across Claude Code sessions.
Takes no parameters.
Example callreturns →
Write or update a Claude Code project memory file (.md). Persists project context, user preferences, feedback, and reference notes across Claude Code sessions. Filename must end in .md with lowercase alphanumeric characters. Path traversal is blocked. Requires authentication.Use this to persist project context, user preferences, feedback, and reference notes across Claude Code sessions. The filena
Takes no parameters.
Example callreturns →
Evaluate a prompt or text against ThinkNEO guardrail policies before sending it to an AI provider. Returns risk assessment, violations found, and recommendations. Requires authentication.
Takes no parameters.
Example callreturns →
Free-tier prompt safety check. Analyzes text for prompt injection patterns and PII (credit card numbers, Brazilian CPF, US SSN, email, phone, passwords). Returns a safety assessment with specific warnings. No authentication required.
Takes no parameters.
Example callreturns →
Check AI governance policies including model access, budget limits, data controls, and agent governance from the ThinkNEO gateway.
Takes no parameters.
Example callreturns →
Check AI budget status including spend vs limit, forecast, and chargeback data from the ThinkNEO gateway.
Takes no parameters.
Example callreturns →
List active alerts for budget, policy, SLA, and security from the ThinkNEO gateway.
Takes no parameters.
Example callreturns →
Get compliance status including framework coverage (EU AI Act, ISO 42001, NIST AI RMF, SOC 2) and governance assessments from the ThinkNEO gateway.
Takes no parameters.
Example callreturns →
Get real-time health and performance status of AI providers routed through the ThinkNEO gateway. Shows latency, error rates, and availability. No authentication required.
Takes no parameters.
Example callreturns →
Schedule a demo or discovery call with the ThinkNEO team. Collects contact information and preferences. No authentication required.
Takes no parameters.
Example callreturns →
Returns usage statistics for your ThinkNEO API key. Shows calls today, this week, this month, monthly limit, remaining calls, top tools used, estimated cost, and current tier. Works without authentication (returns general info).
Takes no parameters.
Example callreturns →
AI Smart Router — find the cheapest model that meets your quality threshold. Specify your task type and quality requirements, and ThinkNEO will recommend the optimal model with estimated cost and savings vs premium models. Supports 17+ models across Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Alibaba, Cohere, and xAI. Requires authentication.
Takes no parameters.
Example callreturns →
Get your AI cost savings report. Shows total requests routed, original cost (what you'd have paid with premium models), actual cost, total savings, savings percentage, breakdown by task type, and model distribution. Requires authentication.
Takes no parameters.
Example callreturns →
Simulate how much your organization would save on AI costs using ThinkNEO Smart Router. Enter your current monthly AI spend and primary model, and see estimated monthly and annual savings with a recommended model mix. No authentication required — try it now!
Takes no parameters.
Example callreturns →
Evaluate your organization AI Trust Score (0-100) across 10 dimensions: Guardrails, PII Protection, Injection Defense, Audit Trail, Compliance, Model Governance, Cost Controls, Outcome Validation, Observability, and Smart Routing. Returns a score, detailed breakdown, badge level (Platinum/Gold/Silver/Bronze/Unrated), and actionable recommendations. Score is valid for 30 days. Generates a public ba
Takes no parameters.
Example callreturns →
Get a public AI Trust Score badge by report token. Returns the organization name, score, badge level, and validity period. Use the badge URL to embed the trust badge in websites and documentation. No authentication required.
Takes no parameters.
Example callreturns →
Search the ThinkNEO MCP Marketplace — the npm for MCP tools. Discover MCP servers and tools by keyword, category, rating, or verified status. Returns name, description, tools count, rating, downloads, and verified badge. No authentication required.
Takes no parameters.
Example callreturns →
Get full details for an MCP server package from the ThinkNEO Marketplace. Returns readme, full tools list, version history, reviews, security score, and installation instructions. No authentication required.
Takes no parameters.
Example callreturns →
Publish an MCP server to the ThinkNEO Marketplace. Validates the endpoint by calling initialize and tools/list, runs automated security scan for secrets and injection patterns, computes a security score (0-100), and stores the entry with version history. Validates the endpoint (calls initialize + tools/list), runs security scan (secrets detection, injection patterns), and stores the entry. Authent
Takes no parameters.
Example callreturns →
Rate and review an MCP server in the ThinkNEO Marketplace. One review per user per package (updates on repeat). Rating from 1 (poor) to 5 (excellent) with optional comment. Reviews affect the package average rating shown in search results. One review per user per package (updates on repeat). Authentication required.
Takes no parameters.
Example callreturns →
Get installation config for an MCP server from the ThinkNEO Marketplace. Returns ready-to-use JSON config for Claude Desktop, Cursor, Windsurf, or custom clients. Tracks the download. No authentication required.
Takes no parameters.
Example callreturns →
Bridge MCP tool registry to A2A format. Shows tool-to-skill mappings.
Takes no parameters.
Example callreturns →
Bridge A2A agents to MCP tool format.
Takes no parameters.
Example callreturns →
Generate an A2A Agent Card from registry data.
Takes no parameters.
Example callreturns →
List all MCP <-> A2A bridge mappings for a tenant.
Takes no parameters.
Example callreturns →
Start a new agent observability trace. Creates a session that tracks all tool calls, model calls, decisions, and errors for an AI agent run. Returns a session_id to use with thinkneo_log_event and thinkneo_end_trace. Requires authentication.
Takes no parameters.
Example callreturns →
Log an event within an active agent trace. Supports event types: tool_call, model_call, decision, error, pii_access, guardrail_triggered. Returns event_id and running session cost. Requires authentication.
Takes no parameters.
Example callreturns →
End an active agent trace and get the session summary. Returns total cost, duration, tool/model call counts, and event count. Triggers post-session anomaly detection (cost spikes, error rate). Requires authentication.
Takes no parameters.
Example callreturns →
Retrieve the full trace for an agent session. Returns the complete timeline of events (tool calls, model calls, decisions, errors), session metadata, total cost, duration, and any alerts triggered. Requires authentication.
Takes no parameters.
Example callreturns →
Get the agent observability dashboard — aggregated metrics for your AI agents. Includes total sessions, events, cost, error rate, latency, top agents, top tools, active alerts, and cost trend over time. Like Datadog, but for AI agents. Requires authentication.
Takes no parameters.
Example callreturns →
Define the pre-AI cost baseline for a business process. Example: 'customer_support_ticket costs $12 per ticket and takes 15 minutes without AI'. This baseline is used to calculate ROI when agents handle the same process. Call this once per process to establish the comparison point.
Takes no parameters.
Example callreturns →
Log a business decision made by an AI agent. Tracks the AI cost and the business value generated. If a baseline exists for the process, value is auto-calculated from the baseline cost. Example: agent 'support-bot' resolved a 'customer_support_ticket' at $0.03 AI cost, replacing a $12 human-handled ticket. ROI: 400:1.
Takes no parameters.
Example callreturns →
Log a risk event that was blocked or avoided by the governance layer. Quantifies the estimated dollar impact of the avoided risk. Examples: PII leak blocked (est. $50K GDPR fine), prompt injection prevented, policy violation caught before production. If estimated_impact_usd is not provided, a default is calculated from severity.
Takes no parameters.
Example callreturns →
Analyze cost-per-decision for AI agents. Shows the actual AI cost for each decision, compared to the pre-AI baseline. Answers: 'How much does each AI decision cost?' and 'How does it compare to doing it without AI?'
Takes no parameters.
Example callreturns →
Calculate ROI per AI agent. Shows value generated vs AI cost consumed, with daily trend, success rate, and comparison to pre-AI baseline. Answers: 'Is this agent generating or consuming value?' and 'What's the ROI trend?'
Takes no parameters.
Example callreturns →
Executive business impact dashboard. Returns a single view of: total value generated by AI agents, total AI cost, net ROI, risk avoided in dollars, cost per decision, top performing agents, and risk event summary. This is the report a CxO needs to justify AI investment.
Takes no parameters.
Example callreturns →
Retrieve A2A (agent-to-agent) interaction logs from the live gateway. Shows which agents called which, actions performed, costs, and outcomes.
Takes no parameters.
Example callreturns →
Retrieve A2A interaction policies from the live gateway. Shows allowed actions, rate limits, cost caps, and approval requirements.
Takes no parameters.
Example callreturns →
Visualize agent-to-agent communication flow. Shows registered agents, their approval status, and interaction patterns from the live gateway.
Takes no parameters.
Example callreturns →
Retrieve immutable audit trail for A2A interactions with hash verification. Each event is cryptographically chained for tamper detection.
Takes no parameters.
Example callreturns →
Detect waste and inefficiency in AI operations. Analyzes agent performance, A2A communication overhead, error costs, unused capacity, and cost outliers. Returns specific actionable findings like 'you are losing $3,200/month on error retries' or 'this flow is 5x more expensive than your best-performing flow'. This is the diagnostic tool that creates the buying trigger.
Takes no parameters.
Example callreturns →
Register an action claim from an AI agent. The agent declares it performed an action (e.g., sent an email, created a PR, wrote a file) and ThinkNEO will verify it actually happened. Returns a claim_id for tracking. Part of the Outcome Validation Loop — 'From Prompt to Proof'. Requires authentication.
Takes no parameters.
Example callreturns →
Trigger verification of a registered action claim. Runs the appropriate verification adapter (HTTP check, file check, database check, etc.) and returns the result with evidence. If already verified, returns cached result (use force=true to re-verify). Part of the Outcome Validation Loop — 'From Prompt to Proof'. Requires authentication.
Takes no parameters.
Example callreturns →
Retrieve the immutable proof record for a verified claim. Includes the original claim, verification evidence, verifier identity, and a SHA-256 proof hash for tamper detection. This is the 'proof' in 'From Prompt to Proof'. Requires authentication.
Takes no parameters.
Example callreturns →
Aggregated outcome verification metrics — verification rates, failure patterns, agent reliability rankings, and daily trends. Shows how reliably your AI agents are delivering verified outcomes. 'Datadog for AI outcomes'. Requires authentication.
Takes no parameters.
Example callreturns →
View the outcome benchmark matrix — real quality scores per provider/model/task_type based on verified outcomes, not static estimates. Shows verification rates, sample counts, and rankings. Requires authentication.
Takes no parameters.
Example callreturns →
Compare providers side-by-side for a specific task type. Shows quality scores, verification rates, and rankings based on real outcomes. Requires authentication.
Takes no parameters.
Example callreturns →
Explain why the Smart Router would choose a specific model for a task type. Shows both benchmark-based (real outcomes) and static quality estimates, and explains the reasoning behind the recommendation. Requires authentication.
Takes no parameters.
Example callreturns →
Generate a compliance report for regulatory frameworks (EU AI Act, ISO 42001, SOC2, NIST). Exports from live audit data.
Takes no parameters.
Example callreturns →
Define or update an SLA (Service Level Agreement) for an AI agent. Set accuracy, quality, cost, safety, or latency thresholds with automatic breach detection and configurable actions (alert, escalate, disable, switch_model). Like SRE SLOs but for AI agent outcomes. Requires authentication.
Takes no parameters.
Example callreturns →
Check current SLA status for all agents or a specific agent. Shows actual metric values vs thresholds, healthy/breached status, and error budget remaining. Automatically records breaches. Requires authentication.
Takes no parameters.
Example callreturns →
View SLA breach history — which SLAs were breached, by which agents, actual vs threshold values, and resolution status. Requires authentication.
Takes no parameters.
Example callreturns →
SLA overview dashboard — all agents, current status, error budgets, and recent breaches (7d). The SRE dashboard for AI agents. Requires authentication.
Takes no parameters.
Example callreturns →
Export audit events from the live gateway. Supports JSON and CSV formats with date range filtering for SIEM integration.
Takes no parameters.
Example callreturns →
Compare available AI models from the live gateway catalog.
Takes no parameters.
Example callreturns →
Estimate token count for text (chars/4 approximation).
Takes no parameters.
Example callreturns →
Analyze prompt and suggest optimizations with live metrics context.
Takes no parameters.
Example callreturns →
Instruct the gateway to rotate an API key.
Takes no parameters.
Example callreturns →
Check connector grants and secrets status from the gateway.
Takes no parameters.
Example callreturns →
Get semantic cache stats from the live gateway runtime metrics.
Takes no parameters.
Example callreturns →
Detect prompt injection attempts in text using guardrail patterns. Also retrieves live guardrails_blocked stats from the gateway.
Takes no parameters.
Example callreturns →
Detect international PII across 30+ document types from 15+ countries: Brazil (CPF, CNPJ, RG, PIS), USA (SSN, EIN, ITIN, Passport), UK (NINO, UTR), Canada (SIN), EU (IBAN, VAT), Germany (Tax-ID), France (INSEE), Spain (DNI/NIE), Italy (Codice Fiscale), Argentina (CUIT), Mexico (CURP/RFC), Australia (TFN/ABN), India (Aadhaar/PAN), China (ID), Japan (My Number), and credit cards (Luhn validated). Re
Takes no parameters.
Example callreturns →
List your configurable Monitor Agent alert rules. Each rule watches a metric (e.g. an API key reaching a % of its budget, workspace spend, error rate) and notifies your channels (email/discord/whatsapp) when it fires. Requires an admin API key.
Takes no parameters.
Example callreturns →
Create a Monitor Agent alert rule. Example: notify by email when API key 'X' reaches 75% of its budget → metric=key_spend_pct_of_budget, operator=gte, threshold=75, scope_value=<api_key_id>, budget_usd=<key budget>. Requires an admin API key.
Takes no parameters.
Example callreturns →
Delete a Monitor Agent alert rule by its rule_id. Requires an admin API key.
Takes no parameters.
Example callreturns →
Send a test notification through a rule's configured channels to confirm delivery (email/discord/whatsapp). Requires an admin API key.
Takes no parameters.
Example callreturns →
Freeyour plan
$0 forever
- Every tool, no call limit
- No card, no account needed
- Source published under a licence you can read
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because nothing bills
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable. Control Plane costs nothing to install and nothing to call, at any volume.
Call allowance · this installenforced at the gateway
Monthly allowance, in calls
—
—
Alert at 80% · then the tool returns a normal MCP error
Used this month: $8.42
$16.58 left
—Grade not published
No rule-by-rule report on file for this release
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
This listing is a hosted endpoint: the publisher runs it and we connect to it. The scanner reads packages and source, and neither exists to read here, so there is no grade — not a withheld one, an unmeasured one. What can be checked instead is on Installation: what it asks to reach and what it writes.
Release history
Pinned to 1.28.0 — the install command below asks for that release. A pin is part of an install, so it is kept for this visit and written down when you install.
1.28.0
undated
installedNo release note was published with this version.
—
no reviews yet
Reviews are earned
Only accounts with at least 50 real tool calls against this server in the last 90 days can post. Ratings are weighted by how much the reviewer actually uses it, and publishers can reply once per review.
Reviews are earned
Writing one takes an account with at least 50 real tool calls against Control Plane in the last 90 days. That is the whole gate — there is no other way to post, which is why the counts beside each review are worth reading.
No reviews yet
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes an account with 50 real tool calls against it.