# Claude Code Skills for 1C: Enterprise: Building and Validating 1C Objects Without the Configurator > A set of 117 skills and 40 rules turns Claude Code into a 1C source-code assembler and validator, working directly on XML exports. The core trade-off: it checks names and structure, not BSL semantics or query plans. **Desko77/claude-code-skills-1c** — 117 скилов и 40 правил для Claude Code: агент собирает исходники 1С (метаданные, формы, расширения, роли, СКД, обработки) из компактного JSON и разбирает обратно. Базовые операции - без Конфигуратора. - Repository: https://github.com/Desko77/claude-code-skills-1c - Stars: 83 · Forks: 17 - Language: Python - License: MIT - Published: 2026-08-26 · Updated: 2026-08-26 · Language: en - Canonical page: https://hysenlabs.com/projects/desko77-claude-code-skills-1c ## What Problem This Solves The skills cover a wide range: configuration-level operations, metadata objects, managed forms, extensions, reports and data processors, database and storage operations, roles, subsystems, layouts, data composition schemas (SKD), XDTO, web publishing, and testing. There are also API references for BSP and ZUP, which let the agent check method calls before writing code. The 40 rules act as persistent constraints across all tasks, covering BSL standards, anti-patterns, EDT requirements, and BSP usage. This is not a single tool but a toolkit, and the size reflects the breadth of 1C's object model. ## How It Works: JSON In, XML Out The format is not guessed from external exports. The README states that the structure is hard-coded in the skill and locked by snapshot tests that compare the output against a reference. This is a significant design choice: it means the generated XML should be stable across runs, and regressions in the output format would be caught by the test suite. The project also includes a configuration index (`1c-config-index`) that builds an index of the entire configuration, which validators use to check an object against the rest of the configuration, not just in isolation. ## Validation Without a Platform However, the README is explicit about the boundary: this is a check of names and structure, not of expression types, BSL semantics, or query plans. Those require the platform. So if your error is a type mismatch in a BSL expression, or a query that runs slowly, these validators will not flag it. That is a real limitation, and the project does not hide it. ## Getting It Running: Two Installation Paths The second route is manual: copy `skills/*` into `~/.claude/skills/`, copy `rules/*.md` into `~/.claude/rules/`, and copy `commands/*` into `~/.claude/commands/`. The README warns that this overwrites any local modifications to the skills, so if you have customized them, you should use the agent-based path, which shows diffs and asks about each divergence. For a first task, you open Claude Code and start with a prompt; the README provides a template for that as well. ## The Skills Catalog: What Is Covered The sheer number, 117 skills, means that the set is not a lightweight add-on. It is a substantial library, and the maintenance burden is on the project maintainers, not the user. The README does not list every skill in the table, but points to a full list below, which is truncated in the provided material. So the exact skill names beyond the examples are not all visible here, but the categories give a clear picture. ## API References: BSP and ZUP The presence of these references also suggests that the project is aimed at developers working with typical configurations, not just custom ones. The BSP is a de facto standard in the 1C ecosystem, and ZUP is a common HR product. If you work without these libraries, the references are irrelevant but not harmful. ## Limitations and Wrong-Tool Cases Finally, the project is young, with the latest release v1.8.1 dated 2026-08-28, and the README is in Russian. The documentation is thorough but in a language that may be a barrier for non-Russian-speaking developers. The 1C ecosystem is predominantly Russian-speaking, so this is not a problem for the target audience, but it is a consideration for others. ## Alternatives and Comparisons The project also has its own validation approach that is more than a simple XML schema check. The configuration index and cross-object validation are a middle ground between pure syntax checking and full semantic analysis. This is a sensible compromise, but it is not a substitute for a real semantic model. If you need that, you would use EDT or the AI-EDT plugin. ## Maintenance and License Upgrade cost is moderate. The agent-based installation path is designed to handle updates without overwriting local changes, as it shows diffs and asks for each divergence. Manual updates, however, would require re-copying and would risk losing customizations. The snapshot tests provide a safety net for the generated output, but they do not cover all possible configurations. Before adopting, you should verify that the generated XML works with your specific platform version, since 1C XML formats can change between releases. The project does not state which platform versions it supports, so this is an open question. ## Conclusion Adopt this if you are a 1C developer who keeps sources in git and wants an agent to handle scaffolding, parsing third-party reports, and configuration export/load without launching the Configurator. Do not adopt it if your work depends on expression type checking, BSL semantics, or query plan analysis, because the validators explicitly do not cover those. Before relying on it, verify that the skill's generated XML matches your platform version's expectations, and test the snapshot-based output against a real configuration. The project's own boundary is clear: it is a name-and-structure checker, not a semantic model. ## FAQ ### What does "Claude code skills" mean? In this set a skill is a ready operation over 1C:Enterprise sources rather than a hint: assemble a metadata object, a form, a role, a composition schema, an extension or an external processing, and parse any of them back. Rules sit next to those operations and set how the agent works, covering BSL standards, antipatterns and EDT requirements. ### Are Claude Code skills free to use? The project is published under the MIT License, with LICENSE.md in the repository root. Nothing in the README describes a paid tier; what the repository ships is skills, rules, commands and scripts, with the AI-EDT plugin kept in a separate repository of its own. ### How many skills and rules does claude-code-skills-1c ship? The repository description says 117 skills and 40 rules, while the group table adds up to 121 and the anchor link to the skills list carries 121 as well. There is also one agent, a 1C scout subagent with restricted tools that the installer places in ~/.claude/agents/. ### What does claude-code-skills-1c need running to validate a change? Neither the 1C platform, nor 1C:EDT, nor a running database. Fourteen validators parse the XML dump, and 1c-config-index compares the object with the rest of the configuration: reference types, form data paths, subsystem membership, role rights, dataset names, table and field names in queries, and common module calls. ### How do I install claude-code-skills-1c without typing commands? Copy the prompt from the README, replace <КАТАЛОГ> with your own directory and give it to the agent in a new session. It clones the repository or runs git pull, reads skills/claude-env-setup/SKILL.md with safety.md and components.md, inventories the machine, shows a plan table and installs only what you choose. ## Sources - [Official README](https://github.com/Desko77/claude-code-skills-1c#readme) - [Project repository](https://github.com/Desko77/claude-code-skills-1c) - [Release notes](https://github.com/Desko77/claude-code-skills-1c/releases) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/desko77-claude-code-skills-1c