Rules\skills\subagens for vibecoding in 1C(bsl)
This repository also configures its own agents. See what ai_rules_1c tells them →
Agent Claude Code
1C business analyst: analyzes existing code and metadata, writes PRDs and specifications, answers architectural questions in 1C terms without writing code. Use PROACTIVELY for requirements analysis or specifications.
Agent Claude Code
Read-only 1C architecture reviewer: confidence-scored review of an existing design — patterns, scalability, 1C best practices. Only on explicit request or pipeline stage 2 with a reviewer model selected per subagents.md; never auto-trigger.
Agent Claude Code
1C solution architect: designs complex modifications — existing patterns, component boundaries, data flows, risks, build sequence. Detailed executable task plans belong to 1c-planner.
Agent Claude Code
Read-only 1C code reviewer: bugs, readability, standards compliance; reports only confidence-filtered important issues. Only when the user explicitly asks for a review and a reviewer model is selected per subagents.md.
Agent Claude Code
1C developer: creates modules, procedures, functions, queries and forms with MCP docs, syntax and metadata checks. Use PROACTIVELY for bulk or multi-module 1C code work; trivial single-file edits stay with the parent (subagents.md).
Agent Claude Code
1C end-user and administrator documentation: user guides, admin manuals, tutorials, codemaps, API references; not inline code comments. Use PROACTIVELY when user-facing documentation must be created or updated.
Agent Claude Code
1C error fixer: resolves syntax errors, runtime errors and BSL Language Server warnings with minimal changes, no architectural modifications. Use PROACTIVELY when errors occur in 1C code.
Agent Claude Code
Read-only 1C codebase explorer, the project's ONLY exploration subagent — use instead of any host built-in Explore / generic scout. MCP-first; levels quick / medium / thorough. Use PROACTIVELY for 'where is X / how does Y work / who calls Z'.
Agent Claude Code
1C metadata specialist: creates, edits, validates and removes configuration objects, forms, SKD, MXL, roles, EPF/ERF, CFE/CF, subsystems, command interfaces. Use PROACTIVELY for any 1C metadata, form, report, layout, role or extension change.
Agent Claude Code
1C performance specialist: finds bottlenecks, optimizes queries and loops, proposes concrete improvements. Use when the user reports slowness, optimization is the explicit task, or a user-requested review found slow code.
Agent Claude Code
1C planning specialist: turns complex features and refactoring into actionable implementation plans — task breakdown, dependencies, risks. Use PROACTIVELY for feature implementation, architectural changes or complex refactoring.
Agent Claude Code
1C refactoring specialist: dead-code cleanup, consolidation, simplification, safe removal of unused code and duplicates, technical-debt reduction. Explicit performance-optimization tasks go to 1c-performance-optimizer.
Agent Claude Code
1C testing agent: deploys via /deploy-and-test and verifies functionality in the 1C interface — QA MCP first, web client as fallback. For UI-test requests or verification under UITESTING, on an authorized dev/test target.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: