1c-agent-skills
An Agent Skills plugin for 1C:Enterprise 8.3 development: core domain rules plus a build-and-database skill, tested across three environments
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- The 1c-build-and-db skill performs a configuration load into a database, an operation with irreversible consequences if a branch is chosen wrong
- It needs read access to the environment's skills directory, which some security configurations close by default
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/1c-agent-skillsDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Run one by one in the Claude Code chat
/plugin marketplace add haflingmax/1c-agent-skills
/plugin install 1c-agent-skills@haflingmax-1cIn Claude Code run /plugin marketplace add haflingmax/1c-agent-skills and /plugin install 1c-agent-skills@haflingmax-1c; in Codex run the equivalent codex plugin marketplace add and codex plugin add, or copy the contents of skills/ into your environment's skills directory by hand.
Other ways from the author
/plugin marketplace add haflingmax/1c-agent-skills
/plugin install 1c-agent-skills@haflingmax-1cThe plugin updates itself when the marketplace updates; no reinstall needed.
This is third-party code. Review the repository files before installing.
What it does
The project builds a skill set in the open Agent Skills format for the 1C 8.3 platform, split not by object type or command but by the developer's area of focus: what am I doing right now (editing a form, writing a query, configuring rights). At publication, two of fifteen planned sections are done: the developing-1c-configurations core with general rules and layout, and 1c-build-and-db for the configuration-load-to-database operation, with branches for file-based vs client-server databases, user lockout, and whether a dynamic update is allowed. The set explicitly excludes specifics of particular standard configurations (ERP, БП, ЗУП) and the БСП library, because their APIs diverge between releases; instead of guessing an API, the core recognizes БСП and reports it as out of scope. Three built-in Python checkers verify skill structure, the sourcing of claims about irreversible operations, and plugin manifest correctness before a commit.
Who it is for. For 1C developers on the 8.3 platform, on any configuration, who care about platform-wide rules and cautious, script-verified database operations rather than guessing a specific standard configuration's API.
Good fit when
- You need 1C 8.3 development rules that hold for any configuration
- You need a safe sequence for loading a configuration into a file-based or client-server database, with key checks
- You use Claude Code, Codex or Kilo Code and want a skill format verified by actual runs, not just documentation
Not a fit when
- You need rules for a specific standard configuration (ERP, БП, ЗУП) or the БСП library: the set explicitly does not cover them and says so honestly
- You need full coverage right away: only two of fifteen planned sections are done
Example request
Load the current configuration into the client-server database with a dynamic update if it is allowedLimitations
Only the core and one applied skill out of fifteen planned sections are ready; the rest are in progress. On Codex, skill auto-loading was not verified on the official model and failed on a minimax-m3 adapter setup (0 of 5 runs). In Kilo Code, the subagent pipeline does not run if the task tool is disabled, and the skills directory needs read permission granted separately. Support in Copilot, Gemini and Cursor is stated only per those environments' documentation and not confirmed by an actual run.
How to disable. In Claude Code or Codex, run the plugin uninstall command (/plugin uninstall or equivalent), or if installed manually, delete the copied folders from the environment's skills directory.
Security check
- The 1c-build-and-db skill performs a configuration load into a database, an operation with irreversible consequences if a branch is chosen wrong
- It needs read access to the environment's skills directory, which some security configurations close by default
README in short
The README explains the scope in detail (the 8.3 platform in general, excluding standard configurations and БСП), the status (two skills done out of fifteen sections), two install paths (plugin via the marketplace or manual copying per environment with a table of verified paths), the knowledge-area-over-object-type layout principle, three working modes (one-off task, platform operation, planning pipeline) and environment quirks including a known Codex issue on minimax-m3. A separate section describes three built-in checkers and the rule that any tool writing to skills/ must end with a green check-manifests.py run. MIT license, content written from scratch against 1C:ITS documentation.
FAQ
Does the set cover БСП or specific standard configurations?
No, this is a deliberate choice: their APIs diverge between releases, so the core recognizes БСП and reports it out of scope instead of inventing rules for it.
How do I check a skill actually loaded in my environment?
The README's table marks where triggering was confirmed by an actual run (Claude Code and Kilo Code via the global path, Codex via acceptance testing) versus only stated per environment documentation.
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