Rentgen: a 1C development environment with MCP
Рентген
Early access to an open platform for 1C development: export snapshots, a dependency graph, a local MCP, and edit drafts with versions and conflict checks
High risk
We rate an entry high when the tool writes to external systems, handles money, production databases or secrets, or runs arbitrary commands. The CLI installs it only with your consent.
Why this level
- The live writer applies supported metadata and BSL changes directly to the source, albeit with an atomic journal and CAS rollback
- The confirmed native business roundtrip scenario includes writing, renaming and deleting data in a database
Install
Manual install
Install the offline core kit per GETTING_STARTED.md into a separate Windows x64 Python 3.11 environment, register the configuration export folder and create the first snapshot, connect the VSCodium/Cline profile from integrations/open-editor, then open modules, save drafts and compare versions.
This is third-party code. Review the repository files before installing.
What it does
Rentgen captures a 1C configuration export into immutable snapshots, builds a dependency graph and impact analysis from them, and stores an agent's proposals as separate drafts with history and version comparison, leaving the source configuration untouched when a draft is created. A local stdio MCP gives an agent project context bounded by an editor profile, and a VSCodium extension shows module and draft trees, search and version comparison. Applying changes has several tiers: workspace apply into an owned copy with inventory checks and rollback, a narrow live writer for one Designer XML layer and catalog attribute renaming, and a separate live writer for one existing BSL file, both with an atomic journal and CAS rollback. A proposal is checked via a local Ollama, BSL diagnostics before and after, the 1C compiler on a separate database built from a saved export, and a trusted YAxUnit profile.
Who it is for. For 1C developers willing to work with an early-access product who want a verifiable, versioned process: from understanding an unfamiliar configuration to carefully applying a change with rollback.
Good fit when
- You want to understand an unfamiliar or complex configuration through a dependency graph and snapshots, without risk to the sources
- You want a local MCP bounded by an editor profile, instead of giving an agent access to the whole project
- You want an agent's proposals kept as drafts with history, version comparison and conflict checks before applying
Not a fit when
- You need a product ready for standard configurations and a live infobase: the README explicitly says these are not yet accepted
- Work isn't on Windows x64 with Python 3.11: that's the currently stated profile
- You need a managed local edit without Ollama: the README requires Ollama with an installed model and a pinned BSL runtime specifically for this scenario
Example request
Show the impact graph for the Article attribute of the Nomenclature catalog and prepare a draft renaming it to SKULimitations
The project is explicitly marked early access: components are tested, but the product is still evolving. Standard configurations and a live infobase are not yet accepted. The native business roundtrip (write, read, rename, delete) is confirmed only on a separate synthetic database, not on an arbitrary configuration. Model verification is limited to one artificial scenario on Qwen3.5:9b and is not a quality assessment for arbitrary 1C tasks. The absence of BSL errors does not confirm business-logic correctness and does not authorize applying a change. The public HTTPS notification service is not deployed, only tested on loopback TLS.
How to disable. Remove the Rentgen extension from VSCodium and stop the local core and MCP processes per GETTING_STARTED.md.
MCP
- Transport
- stdio
- Authentication
- not required
Security check
- The live writer applies supported metadata and BSL changes directly to the source, albeit with an atomic journal and CAS rollback
- The confirmed native business roundtrip scenario includes writing, renaming and deleting data in a database
README in short
The README describes the product's early stage in detail: a table of currently working scenarios (project understanding, MCP context, drafts, tiered change application, editor work, model proposal checking, saved-version checking, archiving, owner reports), a six-step workflow diagram, a verified starting scenario, a roadmap with readiness criteria for the next stages, and a large documentation table covering MCP, workspace, metadata and BSL live apply, owner reports, three-way updates and EDT inventory. It explicitly states the feature list isn't a finished product, and old readiness claims don't apply to the current release. MIT license on the project's own code.
FAQ
Does an agent's draft immediately change the configuration?
No, the README explicitly states the source configuration is not changed when a draft is created; applying it is a separate, further-checked step.
Does the product already work with standard configurations and a production database?
No, the README explicitly states standard configurations and a live infobase are not yet accepted; this is early access.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything