aleksp99/v8std-mcp repository overview

⁠v8std MCP — extended self-contained image

Docker image for the ShadobaAI/v8std⁠ fork of zeegin/v8std⁠.

A self-contained, read-only MCP knowledge server for 1C:Enterprise development, combining v8std standards, corporate development rules, YaXUnit references and Vanessa Automation knowledge.

⁠Included knowledge

The image includes four searchable collections:

⁠Vanessa Automation

The bundled Vanessa knowledge snapshot contains:

• 39 practical guides covering feature files, reusable scenarios, test data, form assertions, synchronization, configuration, execution, debugging and reporting.
• 47 MCP tool reference cards describing documented Vanessa capabilities.
• Source manifests and a source map for tracing the bundled materials.

Vanessa knowledge is available through v8std_search, v8std_get_page and v8std_get_section.

These cards provide documentation. They do not add 47 executable tools to this server or connect it to a running 1C session. To execute Vanessa commands, configure a separate Vanessa MCP connection and use its actual tool schemas.

Resources:

• Vanessa Automation project⁠
• Vanessa Automation documentation⁠
• AI and MCP documentation⁠
• 1C client MCP devkit⁠
• Bundled Vanessa knowledge⁠

The YaXUnit and Vanessa collections are curated references optimized for MCP retrieval, rather than complete copies of their official documentation.

⁠Differences from zeegin/v8std

This fork adds:

• Additional development rules from docs/corporate⁠.
• YaXUnit knowledge from docs/yaxunit⁠.
• Vanessa Automation knowledge from docs/vanessa⁠.
• 11 MCP tools, including compact summaries, sections, YaXUnit pattern/API cards and contextual requirements.
• Collection filters for searching a specific knowledge source.
• Internal v8std: references for navigating the corpus through MCP.
• A built-in knowledge index, generated during Docker build.

The container starts with the corpus included in the image. Normal operation requires no repository checkout, knowledge volumes or public snapshot download.

The image contains the MCP service; the documentation website is not included.

⁠MCP tools

Search supports continuation through next_cursor. Continue with the same query, filters, mode and limit.

To search Vanessa guides, for example:

{
  "query": "подготовка данных повторный запуск",
  "collections": ["vanessa"],
  "types": ["pattern"]
}

Pass returned references such as v8std:vanessa:patterns:project-layout to the reading tools.

The server retrieves reference material and provides heuristic snippet guidance. It does not inspect entire projects, run analyzers, execute tests or modify source code.

⁠Run

docker pull aleksp99/v8std-mcp:latest

docker run --name v8std --restart unless-stopped -p 127.0.0.1:8766:8766 -d aleksp99/v8std-mcp:latest

Endpoints:

The example exposes the service on the Docker host's loopback interface.

⁠Update an existing container

Pulling an image does not replace an existing container. To apply an update:

docker pull aleksp99/v8std-mcp:latest
docker stop v8std
docker rm v8std
docker run --name v8std --restart unless-stopped -p 127.0.0.1:8766:8766 -d aleksp99/v8std-mcp:latest

⁠Client configuration

Configure your MCP client to use Streamable HTTP at:

http://127.0.0.1:8766/mcp

Configuration keys depend on the client. For clients accepting the mcpServers / url format:

{
  "mcpServers": {
    "v8std": {
      "url": "http://127.0.0.1:8766/mcp"
    }
  }
}

⁠Image properties

• Python 3.12 runtime.
• Non-root user: UID/GID 10001.
• Built-in Docker HEALTHCHECK.
• Default port: 8766.
• Default MCP path: /mcp.
• Build-time generation of pages.jsonl, search-vectors.jsonl, llms.txt and llms-full.txt.
• Built-in YaXUnit and Vanessa manifests, Vanessa source map and license texts.
• No knowledge-base download or index generation required at startup.

The bundled knowledge changes when the image is rebuilt. The container does not automatically synchronize external documentation.

⁠Build and publication

Maintainers build the image manually from the fork checkout using:

build-v8std-mcp.cmd

The script creates a Docker Hub tag using the configured namespace. Publication is a separate manual docker push.

⁠Example: AI development harness

The Kafka Tools AI harness⁠ demonstrates how v8std is used in an AI-assisted 1C development workflow.

The harness combines project code search, local development tools and v8std reference knowledge. It:

• retrieves applicable development standards and corporate rules before making BSL changes;
• selects relevant requirements according to the task and affected mechanisms;
• loads specific sections through v8std_get_section and compact summaries through v8std_get_summary;
• references centrally maintained rules instead of duplicating their full text in agent instructions.

Project routing, skills and executable checks belong to the harness. v8std supplies the reference knowledge used during planning, implementation and review.

See the harness README⁠ for installation instructions and workspace requirements.

⁠Source and attribution

• ShadobaAI/v8std fork⁠
• zeegin/v8std upstream project⁠
• Additional development rules⁠
• Bundled YaXUnit knowledge⁠
• YaXUnit project⁠
• Bundled Vanessa knowledge⁠

This image is based on v8std and includes modifications maintained in the ShadobaAI/v8std fork.

YaXUnit materials originate from the bia-technologies/yaxunit project.

Vanessa materials are adapted from the Pr-Mex/vanessa-automation project. The MCP reference also uses documentation from 1c-neurofish/onec-client-mcp-devkit.

Bundled license texts are available under /opt/v8std/LICENSE and /opt/v8std/LICENSES. See the corresponding source repositories for complete attribution and licensing information.