Конвейер поиска и реализации через цепочку MCP-инструментов

orchestration

mcp-chaining is a Claude Code skill that implements a research-to-implement pipeline chaining 5 MCP tools with graceful degradation across each step. The pipeline connects nia for documentation search, ast-grep for structural code pattern matching, morph for fast codebase search (20x faster than grep), qlty for code quality validation, and git for version control operations. If any tool is unavailable or missing an API key, the pipeline skips that step and continues rather than failing entirely — each step returns a typed StepResult with SUCCESS, SKIPPED, or FAILED status. Core files include `scripts/research_implement_pipeline.py` for the main implementation and `scripts/test_research_pipeline.py` for running tests in an isolated sandbox. The skill targets developers building multi-tool MCP workflows, debugging subprocess environment variable propagation, or looking for a reusable pipeline template to extend with their own steps.