treadle
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@treadleRename task 'Approve' to 'Review' in order.bpmn"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Treadle
Read, explain, lint and edit the .bpmn files already in your repo — from any AI agent,
without wrecking the diagram.
An open-source MCP server, CLI and library for BPMN 2.0. Engine-neutral, file-native, no account, no server, works offline. Apache-2.0.
Status: week 0. No product code yet. This repo currently holds the benchmark harness that decides how the product gets built. See docs/FINDINGS.md for what has been measured and docs/DECISIONS.md for what was decided on the strength of it.
Why
Analysts inherit BPMN models far more often than they draw them. Every AI tool in this space generates new diagrams: Camunda's Copilot "officially supports only modifying diagrams that were created by the BPMN Copilot itself", and every open-source BPMN MCP server is a text-to-diagram generator abandoned within days of creation.
Meanwhile, editing a real .bpmn file with ordinary text tools breaks in structural ways:
a flow id appears three times in the XML (
sequenceFlow,incoming,outgoing), so a string replace on it is ambiguous by construction and silently corrupts adjacencypast ~60 nodes the file exceeds an agent's tool-response budget, so read-modify-write stops working at all
regenerating the file moves every shape, producing a diff nobody will review
Treadle operates on the parsed document instead: patches apply to the object tree, and export rewrites only what changed.
Related MCP server: MCP-BPMN Server
What is measured, not claimed
One-attribute edit on a normalized 1,710-line file | −1 +1 lines, 0 shapes moved |
Camunda 8 / Zeebe extension round-trip | lossless, 15/15 probes, 0 warnings |
MIWG reference models parsed | 22/22, 0 errors |
MIWG reference models XSD-valid | 22/22 |
| fails on 9/22 — 8 silent, 1 crash |
That last row is why this is an editing tool and not a diagram generator. Details and repro steps: docs/FINDINGS.md.
Repo layout
bench/ the week-1 bake-off harness
corpus/ BPMN fixtures + profilers (see corpus/PROVENANCE.md)
scorer/ the five scoring gates
tasks/ edit tasks and their assertions
arms/ A naive · B strong baseline · C IR prototype
third_party/ OMG BPMN 2.0 XSD schemas, unmodified
docs/ findings, decisionsRunning the harness
Needs Node 22.12+ (why).
npm install
node bench/corpus/profile.mjs bench/corpus
node bench/corpus/probe-diff-floor.mjs bench/corpus
node bench/scorer/probe-layout.mjs bench/corpus
node bench/scorer/baseline.mjs bench/corpusThe bake-off
Before any product code, three arms are scored on the same tasks by the same gates:
A — naive. Read the whole file, edit with find-and-replace, write back.
B — strong baseline. Edit the XML directly, then post-process with bpmnlint and auto-layout, feeding lint errors back for one repair turn. Roughly one week of work. This is the arm to beat. If it wins, it is the product.
C — structured. Parse to a tree, project a compact view, apply patch operations, re-serialize.
Decision rules are committed in advance in docs/DECISIONS.md.
License
Apache-2.0. Contributions by DCO sign-off — no CLA, and no plan to relicense the core.
Third-party components and fixture provenance are recorded in NOTICE and
bench/corpus/PROVENANCE.md. Treadle's own packages depend
only on OSI-licensed software; bpmn-js and the other bpmn.io watermark-licensed
toolkits are excluded from the core by CI policy
(ADR-009).
BPMN is a trademark of the Object Management Group. Camunda, Signavio and ARIS are trademarks of their respective owners. Treadle is not affiliated with or endorsed by any of them.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Connectors
Versioned artifact review for people and AI agents, with contextual comments and human control.
Versioned documentation registry and semantic search for AI tools and coding assistants.
Git-backed platform for skills, tools, and context for AI agents
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to manage OSOP workflows by providing tools for validation, execution, testing, and optimization. It also supports converting workflows between OSOP and external formats like GitHub Actions, BPMN, and Airflow DAGs.1Apache 2.0
- FlicenseBqualityDmaintenanceEnables AI agents to create, manipulate, and manage BPMN 2.0 diagrams programmatically, with support for Mermaid conversion, auto-layout, and file persistence.249
- FlicenseNot gradedqualityFmaintenanceEnables AI-driven graphical diagram creation and manipulation using natural language, with support for BPMN workflows, analysis, and manual editing via the Model Context Protocol.1
- AlicenseBqualityBmaintenanceEnables AI assistants to create and manage one BPMN 2.0 diagram at a time, including Mermaid conversion, validation, layout, persistence, and XML or SVG export.27MIT
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/EduardoHOS/treadle'
If you have feedback or need assistance with the MCP directory API, please join our Discord server