edi-mcp
Let AI agents read, validate and acknowledge EDI documents. Parses raw X12 and EDIFACT interchanges into structured JSON, validates envelope integrity, produces plain-language summaries, and generates 997 Functional Acknowledgments.
README
edi-mcp
Let AI agents read, validate and acknowledge EDI documents.
An MCP (Model Context Protocol) server that parses raw X12 and EDIFACT interchanges into structured JSON, validates envelope integrity, produces plain-language summaries, and generates 997 Functional Acknowledgments. Zero runtime dependencies beyond the MCP SDK - the tokenizer is hand-rolled and reads delimiters from the wire, the way real-world files demand.
Works with Claude (Desktop, Code, API), Cursor, and any MCP-compatible client.
Why
EDI predates JSON by decades and still moves most B2B commerce: purchase orders (850/ORDERS), invoices (810/INVOIC), ship notices (856). None of it is reachable by no-code AI connectors - there is no REST API to point Zapier at, just structured text flowing over AS2/SFTP. This server gives an AI agent eyes on that traffic:
- "Summarize the PO that just landed in the inbox folder"
- "Does this 810 match the PO number and line totals we sent?"
- "Draft the 997 acknowledgment for this interchange"
Tools
| Tool | What it does |
|---|---|
parse_edi |
Raw X12/EDIFACT (auto-detected) → structured JSON: envelope, groups, transactions, segments |
validate_edi |
Envelope integrity: control number agreement (ISA/IEA, GS/GE, ST/SE, UNB/UNZ, UNH/UNT) + declared counts |
summarize_edi |
Plain-language business summary. Deep support: X12 850/810/856, EDIFACT ORDERS/INVOIC. Other types get a segment inventory |
generate_997_ack |
X12 997 Functional Acknowledgment answering an interchange, sender/receiver swapped, AK2/AK5 per transaction |
Quick start
git clone https://github.com/heocoi/edi-mcp.git
cd edi-mcp
npm install && npm run build
Claude Code:
claude mcp add edi -- node /path/to/edi-mcp/dist/index.js
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"edi": {
"command": "node",
"args": ["/path/to/edi-mcp/dist/index.js"]
}
}
}
Then try: "Parse samples/850-purchase-order.edi and tell me the PO number, ship-to address and order value."
Example
Input (X12 850, abbreviated):
ISA*00*...*ZZ*ACMECORP*ZZ*WIDGETSUPPLY*260719*1030*U*00401*000000101*0*P*>~
GS*PO*ACMECORP*WIDGETSUPPLY*20260719*1030*101*X*004010~
ST*850*0001~
BEG*00*SA*PO-2026-4521**20260719~
PO1*1*48*EA*9.75**VP*WS-4417*UP*012345678905~
...
summarize_edi output:
X12 interchange 000000101 - ACMECORP → WIDGETSUPPLY, dated 2026-07-19
## 850 Purchase Order (control 0001)
- Purchase Order PO-2026-4521 dated 2026-07-19 (type SA)
- Ship-to: Acme Corp Warehouse 7
- Line 1: 48 EA @ 9.75 - VP WS-4417, UP 012345678905
- Line 2: 12 CS @ 142.00 - VP WS-9902
- Computed order value: 2172.00
- CTT declares 2 line item(s)
Design notes
- Delimiters come from the wire, not from config. X12 delimiters are read from the ISA segment itself; EDIFACT service characters from UNA when present, standard defaults otherwise. EDIFACT release-character escaping (
?+→ literal+) is honored. - Lenient parse, strict validate.
parse_edirecovers what it can and reports anomalies as warnings;validate_ediis the strict pass. Real-world EDI is messy - a parser that throws on the first oddity is useless for triage. - Acknowledgment policy is explicit.
decideAckStatus()insrc/ack997.tsmaps validation results to A/E/R. Default: clean → A, structural errors → R. Trading partner agreements differ; the policy is one small function on purpose.
Scope and roadmap
Current scope is read/triage/acknowledge - the 80% of daily EDI pain that needs no certification. Not yet included: 999 Implementation Acknowledgment, transaction-set-level schema validation (dictionaries per version), generating outbound 810/856, AS2 transport. Tracking the MCP 2026-07-28 spec revision.
Custom integrations
I build custom MCP servers for systems that no off-the-shelf connector covers: legacy ERPs, proprietary internal tools, SOAP/EDI/flat-file pipelines. If your business runs on a system your AI tools can't see, reach out: anhphong.pham@gmail.com · anhphong.dev
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.