Package Exports

@doquflow/server/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@doquflow/server) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

@doquflow/server

MCP server that lets AI agents read any codebase and build a persistent, incrementally-maintained knowledge base using the LLM Wiki pattern. 15 tools. Zero AI inside — the agent does all the thinking.

AI Agent (Claude, Copilot, Cursor)
       │ calls MCP tools
@doquflow/server          ← reads files, builds wiki, answers questions
       │ filesystem only
Any project on disk       ← any language, any structure

Install

npx @doquflow/cli init

The CLI registers this server automatically. You do not need to install @doquflow/server directly.

Tools (15 total)

Code Extraction

Tool	Input	Output
`read_module`	`{ path }`	Language, classes, functions, dependencies, DB tables, endpoints, config refs, raw content
`list_modules`	`{ path, extensions? }`	All modules in a directory tree (bulk extraction, no raw content)
`write_spec`	`{ project_path, filename, content }`	Writes markdown spec to `.docuflow/specs/`
`read_specs`	`{ project_path, module_name? }`	Reads existing specs; includes `stale: boolean` per spec

Wiki Pipeline

Tool	What it does
`ingest_source`	Parse source doc → extract entities/concepts → create wiki pages with context
`update_index`	Regenerate `index.md` and append to `log.md`
`list_wiki`	List wiki pages by category; includes `stale: boolean` per page and `stale_pages` count
`wiki_search`	BM25-inspired relevance search across all wiki pages
`query_wiki`	Main interface: search + synthesize answer from wiki
`answer_synthesis`	Build structured markdown answer from selected pages
`save_answer_as_page`	Save synthesised answer back to wiki

Health & Guidance

Tool	What it does
`lint_wiki`	Health checks: orphan pages, stale content, broken links, metadata gaps, contradictions. Returns 0-100 health score.
`get_schema_guidance`	Detect domain → recommend missing wiki pages
`preview_generation`	Show what a tool will do before running (reads real wiki state)

Dependency Analysis

Tool	What it does
`generate_dependency_graph`	Build import/shared-table/shared-endpoint graph. Returns `nodes`, `edges`, `most_connected` top 10 (highest-risk files).

Languages supported

TypeScript, JavaScript, Python, Go (structs, funcs, gin/gorilla routes, GORM), Ruby/Rails (classes, ActiveRecord, Rails routes), Rust, Java, C#, PHP, Kotlin, Swift, SQL, Shell, YAML, JSON, and more.

Unknown file types return full raw content — never fails on unfamiliar files.

Extraction engine

Field	Detected from
`classes`	`class`, `interface`, `struct`, `enum`, `record`, Go `type … struct/interface`, Ruby `module`
`functions`	Keyword-prefixed declarations, arrow functions, Go `func`, Ruby `def`
`dependencies`	`import`, `require()`, Go import blocks, Ruby require
`db_tables`	SQL `FROM/JOIN/INTO`, EF `DbSet<T>`, GORM `db.Table()`, ActiveRecord associations
`endpoints`	.NET attributes, Express/NestJS, gin/gorilla/chi routes, Rails route helpers
`config_refs`	`process.env`, `os.Getenv`, `ENV['KEY']`, `IConfiguration`, `appsettings`

Requirements

Node.js 18+
No API keys, no network calls, no AI dependencies

License

MIT — github.com/doquflows/docuflow