CodePilot1C for 1C:EDT

CodePilot1C (EDT plugin)

A 1C:EDT plugin with a built-in MCP Host: an agent gets typing, metadata and the EDT model directly, not by parsing XML files

PluginEditors’ pick

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 MCP Host can be configured to auto-allow all mutating tools with no confirmation
  • Access includes a runtime debugger and profiler that can affect configuration execution
All reasons and checks
Russian stack

ondysss/codepilot1c-edt

Install

Manual install

claude mcp add --transport http -s user codepilot1c http://127.0.0.1:8765/mcp

After enabling the MCP Host in the plugin settings or via 1cedt.ini flags.

This is third-party code. Review the repository files before installing.

What it does

CodePilot1C installs as a regular Eclipse RCP/OSGi plugin in 1C:EDT and starts an inbound MCP Host over HTTP on port 8765, with OAuth 2.1 authorization or a fallback Bearer token. Claude Code, Cursor and Codex agents connect to this endpoint and get capabilities the plugin builds through its understanding of the EDT model rather than by textually parsing files: BSL typing, metadata, forms, SKD, roles, diagnostics, a debugger and runtime profiler, running YAxUnit. Separately there is a standalone CLI with an interactive shell (java -jar codepilot-cli.jar shell) that can run through EDT's active LLM provider (connected mode) or its own OpenAI-compatible provider (standalone mode), asks for confirmation on risky and mutating MCP tools, and stores redacted session transcripts locally.

Who it is for. For 1C developers who work in EDT and want an agent to understand the configuration model as precisely as EDT itself does, not approximately from files.

Good fit when

  • Development happens in 1C:EDT and you need agent MCP access to typing, metadata, forms and SKD
  • You need to run YAxUnit tests and the profiler through an agent
  • You want to gate mutating agent actions behind explicit confirmation via the interactive shell

Not a fit when

  • Development happens in the Configurator, not EDT: the plugin is built specifically for EDT
  • You plan to embed or redistribute the plugin's code in a closed product: the AGPL-3.0 license requires source disclosure
  • You lack JDK 17+ to build or run the CLI

Example request

Use CodePilot1C to find every place a deprecated method is called and show them with precise parameter typing

Limitations

The AGPL-3.0 license imposes copyleft requirements on derivatives and network use. The README's example for auto-starting the MCP Host sets defaultMutationDecision=ALLOW and exposedTools=*, which permits all mutating tools with no confirmation if configured that way; a narrower policy is advisable for safe use. Updating the plugin between incompatible versions requires a manual p2 director procedure with backups, and direct reinstallation may fail to resolve due to singleton bundles.

How to disable. Disable the MCP Host in Preferences, 1C Copilot, MCP Host, or remove the plugin via Help, About 1C:EDT, Installation Details, Uninstall.

Security check

  • The MCP Host can be configured to auto-allow all mutating tools with no confirmation
  • Access includes a runtime debugger and profiler that can affect configuration execution

README in short

The README describes installation via an update site or an offline ZIP, a detailed procedure for updating between incompatible versions through p2 director with backups and diagnostics, the interactive CLI shell with connected/standalone/auto modes and a confirmation system for mutating tools, building via Maven with Tycho, the inbound MCP Host with OAuth 2.1 and auto-start via 1cedt.ini, configuration examples for Claude Code, Cursor and Codex, and links to Infostart articles on using CodePilot1C and the Qwen Code CLI.

FAQ

Is a separate LLM API key needed for the CLI?

Not necessarily: connected mode uses the LLM provider already configured in EDT through a broker, without reading its key directly.

Can dangerous agent actions be blocked by default?

Yes, the interactive shell asks for confirmation on risky and mutating tools unless the defaultMutationDecision policy is explicitly set to ALLOW.

Editors’ pick

A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review

PluginMedium riskNo VPN needed292.5KRepository stars
Editors’ pick

Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture

SkillLow risk271.4KRepository stars
Editors’ pick

GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation

CLIMedium riskNo VPN needed139.3KRepository stars

Reference MCP servers

Model Context Protocol servers

Official

Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything

MCP serverMedium risk90.6KRepository stars
Foxx AICodePilot1C for 1C:EDT

I am Foxx AI and I have already vetted this tool. Ask about install, setup or anything else, and I will keep it simple.