MCP

NaN-EcoNet EcoPass MCP: A First-Attempt Guide to Querying EPR Audit Data Safely with AI Agents

If you want AI agents to query EPR audit data with whitelist-guarded Text-to-SQL, EcoPass MCP is the entry point; but it is part of a three-subsystem monorepo, so clone the whole repository and confirm ports and configuration before starting.

Official source · NaN-EcoNet

Install

git clone https://github.com/chinhanxt/NaN-EcoNet.git && make install-all

Who This Is For

NaN-EcoNet is a monorepo for the green circular economy. The MCP-related part is EcoPass's Enterprise BI Copilot & MCP Tools: it implements the Model Context Protocol, allowing AI agents to discover tools, execute safe Text-to-SQL protected by multi-layer AST and whitelist guardrails, and retrieve EPR audit data according to JSON Schema v1.

  • Good fit: teams that need AI agents to query enterprise EPR (Extended Producer Responsibility) audit data via MCP, with data Q&A powered by guardrail-protected Text-to-SQL.
  • Not a fit / prerequisites: the project targets a specific environmental recycling business scenario and includes three subsystems — Smart Collection Scheduling, Citizen Bulky-Waste Recycling, and EcoPass Enterprise & MCP. If you only need a standalone general-purpose MCP server, the official evidence does not state whether the MCP portion can be deployed separately; this is unknown.
  • The project is labeled Apache-2.0 licensed, v1.0.0; beyond that, no minimum runtime environment requirements are given in the evidence.

First Attempt in 3 Minutes

The official primary path is to clone the repository and install using the root Makefile:

git clone https://github.com/chinhanxt/NaN-EcoNet.git
make install-all

make install-all installs dependencies for all three subsystems; this is the recommended installation method in the official documentation.

After installation, start the EcoPass Enterprise services and MCP (official ports are 3010/3011/5002):

make dev-ecopass

Success signals: the EcoPass Enterprise and MCP-related ports (3010, 3011, 5002) are up and reachable; the MCP service provides tool discovery via the MCP protocol on port 3011.

The official documentation also offers an alternative way to start everything: make docker-up, which brings up all services at once using deploy/docker-compose.yml. If you only want to run the MCP-related parts first, make dev-ecopass is the more direct path.

For the first attempt, stick to low-risk actions: have your MCP client call tool discovery (list tools), or run a read-only SELECT-type Text-to-SQL query, and observe the returned tool list and query results. Do not perform any write or commit operations on your first attempt. Even with guardrails, the official README excerpt does not specify which statements are allowed through or whether additional credentials are required for Text-to-SQL — refer to the documentation in the repository.

Use Cases

Use case 1: Have an AI agent discover available tools and retrieve EPR audit data.

Scenario: you have started the EcoPass MCP (port 3011). How to call it: connect from your MCP client, use MCP tool discovery to list the tools the server provides, then invoke the tool for retrieving EPR audit data. Expected result: the server returns EPR audit data according to JSON Schema v1. The official evidence confirms these capabilities exist, but does not provide specific tool names or parameter lists — rely on the tool list returned by the server at runtime.

Use case 2: Run a data Q&A with guardrail-protected Text-to-SQL.

Scenario: you want to query business data in natural language. How to call it: invoke its Text-to-SQL tool via MCP. This capability has built-in multi-layer AST and whitelist guardrails, which the official docs describe as "safe Text-to-SQL". Expected result: a restricted SQL query is generated and executed, returning query results. Note: the specific whitelist rules of the guardrails, the allowed SQL dialect, and required credentials are not stated in the official evidence; if a query is blocked by the guardrails, that is expected behavior. Start with read-only SELECT statements to verify.

Boundaries

  • Read/write nature: the EcoPass MCP provides tool discovery, Text-to-SQL, and EPR audit data retrieval. The official summary calls the Text-to-SQL "safe" with multi-layer AST and whitelist guardrails, but that only means statement execution is restricted — it does not mean all operations are side-effect free. Whether write operations exist and how they are confirmed is not stated in the evidence.
  • Credentials and local requirements: the official README excerpt does not list required API keys, database credentials, or environment variable configuration; consult the repository documentation before running.
  • Ports: EcoPass Enterprise & MCP use 3010/3011/5002; the Smart Collection Engine uses 8000/8501/8502; the Citizen Bulky Portal uses 3006.
  • Officially stated restrictions: Text-to-SQL is subject to multi-layer AST and whitelist security guardrails; EPR data is returned according to JSON Schema v1.
  • Key information not yet confirmed: which MCP clients the server supports, whether there is an officially hosted remote deployment option, the exact paths of configuration files, and whether running only the MCP portion is officially supported — none of these are documented.

These unconfirmed items do not mean there is no risk. Before reading the full documentation in the repository, do not make connecting to a production database or real EPR data sources your first attempt.