DeHor-Labs/mcp-fiscal-brasil

Description and use

**Brief Summary**
DeHor‑Labs / mcp‑fiscal‑brasil is a Python‑based open‑source server that implements Brazil’s MCP fiscal protocols (CNPJ, NF‑e, NFS‑e, CT‑e, SPED, eSocial, Simples Nacional, Reforma 2026). It bundles 44 ready‑to‑use tools, works offline with pre‑loaded tables, and requires no registration, making it a turnkey backend for fiscal data integration.

**Value**
The project abstracts the complex, government‑mandated fiscal APIs into a single, standard‑compliant MCP endpoint, enabling AI assistants, RPA bots, or custom applications to query and manipulate real‑time tax, invoice, and social‑security data without dealing with the myriad of native Brazilian services. By exposing a uniform Model Context Protocol (MCP) interface, it turns disparate fiscal systems into consumable services, accelerating development of compliance‑aware AI agents and reducing integration overhead for fintechs, ERP vendors, and government portals.

**Practical Adoption Path**

| Step | Action | Why |
|------|--------|-----|
| 1️⃣ | **Clone & install** – pip install -r requirements.txt && python -m mcp_fiscal_brasil | Quick local spin‑up; no external registration needed. |
| 2️⃣ | **Load offline tables** – run the provided load_tables.py script to populate the embedded SQLite/PostgreSQL database with CNPJ, IBGE, tax‑rate, and schema data. | Guarantees deterministic behavior and compliance with the latest fiscal rules. |
| 3️⃣ | **Configure endpoints** – edit config.yaml to enable only the required modules (e.g., NF‑e, SPED). | Minimizes attack surface and resource usage. |
| 4️⃣ | **Integrate via MCP SDK/CLI** – use the supplied Python SDK (mcp_fiscal_brasil.client) or the CLI (mcp-cli) to call methods such as get_cnpj, validate_nfe, submit_esocial. | Provides a language‑agnostic contract; other languages can call the REST/GRPC layer. |
| 5️⃣ | **Connect AI agents** – point your LLM‑based assistant (e.g., LangChain, AutoGPT) to the MCP endpoint, using the standard tool schema. | Enables the assistant to retrieve real fiscal data and perform compliant actions on behalf of users. |
| 6️⃣ | **Production hardening** – enable TLS, set up a reverse proxy, add API‑key authentication, and configure periodic table updates from the official government feeds. | Aligns the service with enterprise security and compliance policies. |

**Production Readiness**

• **Activity & Community** – 121 ★, 24 forks, last commit on 2026‑07‑05; active issue discussion and PR turnover.
• **Stability** – Core fiscal modules are covered by unit tests; the server runs in Docker and Helm charts are available for Kubernetes deployments.
• **Security** – No known CVEs; the repository includes a basic security policy and CI scanning, but a formal audit of the license (MIT‑style) and dependency tree is still recommended.
• **Scalability** – Stateless API layer plus optional PostgreSQL backend allows horizontal scaling; offline tables keep latency low even under high load.

Overall, mcp‑fiscal‑brasil is mature enough for a pilot in regulated environments and can be promoted to production after the usual hardening steps (TLS, auth, dependency audit).

Use in a workflow

DeHor-Labs/mcp-fiscal-brasil helps connect AI assistants to real tools and data through a standard protocol.

• connect AI agents to tools
• ship Model Context Protocol servers
• standardize integrations

Limitations and review notes

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.