KnowledgeRail

KnowledgeRail is a local-first MCP server that turns project documentation and source code into durable, evidence-backed context for AI agents.

It is designed for agents that need to understand, change, review, or document a codebase without loading the whole repository into the model context. Retrieval is bounded, provenance is preserved, missing evidence is reported explicitly, and difficult queries widen progressively instead of silently losing relevant information.

Current status: initial public release 1.0.0. The server uses MCP SDK 2.x and protocol 2026-07-28. The supported transport is local stdio; a remote/serverless deployment is not implemented yet. See SERVERLESS.md.

What it provides

• A menu-first workflow that guides agents through reading, ingestion, code evidence, document generation, and administration.

• Task-aware hybrid retrieval with lexical, graph, passage, and optional semantic evidence.

• Progressive widening with explicit coverage signals and GAP/unknown reporting.

• Complete source ingestion through bounded segments, a coverage ledger, and durable Evidence IR.

• A deterministic TypeScript/JavaScript code index with symbol and reference lookup.

• Incremental graph, retrieval, and semantic indexes stored beside the project wiki.

• Contract-driven Markdown deliverables and gated DOCX export with Mermaid diagrams.

• Conservative migration of existing v1/v2/v3 wikis.

KnowledgeRail does not call an LLM itself. The connected MCP client chooses and calls the tools. OCR and embeddings are optional external providers configured by the user.

Related MCP server: memory-mcp

Requirements

• Node.js 22.12.0 or newer

• npm

• macOS, Windows, or Linux

Mermaid CLI and its compatible Chromium runtime are installed with the package. A separate global Mermaid installation is not required.

Install and run

From source

git clone https://github.com/Deviank88/KnowledgeRail.git
cd KnowledgeRail
npm ci
npm run build

Start it from the project whose knowledge you want to manage:

cd /path/to/your-project
node /absolute/path/to/KnowledgeRail/dist/index.js

From npm

After the first public package release:

npm install --global knowledge-rail
cd /path/to/your-project
knowledge-rail

The npm package is not considered available until it is published by the maintainers. Installing from source is the supported pre-release path.

MCP client configuration

Use the standard stdio server shape supported by your MCP client:

{
  "mcpServers": {
    "knowledge-rail": {
      "command": "knowledge-rail",
      "env": {
        "WIKI_ROOT": "/absolute/path/to/your-project"
      }
    }
  }
}

For a source checkout, replace knowledge-rail with Node and the compiled entry point:

{
  "mcpServers": {
    "knowledge-rail": {
      "command": "node",
      "args": ["/absolute/path/to/KnowledgeRail/dist/index.js"],
      "env": {
        "WIKI_ROOT": "/absolute/path/to/your-project"
      }
    }
  }
}

Client configuration wrappers and file locations differ, but the command, args, and env values are portable. Use absolute paths on every operating system. When WIKI_ROOT is omitted, KnowledgeRail uses the server process working directory. Legacy MCP clients may provide Roots; modern 2026-07-28 sessions do not need them.

Agent workflow

Every MCP 2.0 task starts with:

knowledge_menu {}

The menu returns five areas:

Call knowledge_menu again with the selected area, choose one returned operation, execute only the next action, and report its observed outcome back to the menu. This keeps the workflow discoverable without profiles or hidden client configuration.

A normal context request is:

knowledge_menu {"area":"read","operation":"understand"}
knowledge_context {
  "intent":"understand",
  "objective":"Explain how lease renewal and expiry work",
  "response_detail":"compact",
  "heuristic_token_budget":2000
}

On MCP 2026-07-28, knowledge_context returns selected knowledge-rail:// resource links. The client should materialize only the passages it needs with resources/read, then report coverageSufficient and evidenceGaps back to knowledge_menu.

Why context has a token budget

The budget bounds evidence sent to the model; it does not declare omitted knowledge irrelevant. If coverage is insufficient because of the budget, the guided read workflow widens from 2,000 to 4,000, 8,000, and at most 12,000 heuristic tokens. If evidence is still missing, the result must expose a gap rather than invent an answer.

response_detail="compact" is recommended for normal agent use. full keeps the complete historical TaskContext payload for diagnostics and integrations that need it.

Project data

knowledge_init creates this structure inside the selected project. The roots are intentionally stable: wiki/ is canonical agent memory; docs/ is the document plane for sources, normalized copies, durable evidence state, and deliverables.

project/
├── wiki/
│   ├── index.md
│   ├── log.md
│   ├── SCHEMA.md
│   ├── .knowledge-rail/     # derived indexes, manifests and migration state
│   └── <page-type>/         # created lazily when the first typed page is written
└── docs/
    ├── client/
    ├── transcripts/
    ├── reports/
    ├── changelogs/
    ├── normalized/
    ├── evidence-ir/         # durable Evidence IR and knowledge-recovery state
    ├── deliverables/
    └── assets/

Markdown pages are canonical knowledge. Files below wiki/.knowledge-rail/ are derived or operational state and can be rebuilt where the corresponding workflow supports it. Source documents remain under docs/; normalization never overwrites the original.

These directories may contain private project information. Decide deliberately whether the consuming project should commit them.

Document memory and deliverables

Document generation starts with knowledge_menu {"area":"document","operation":"create"}. The guided flow calls knowledge_plan_document, compiles a separate bounded evidence pack for every section, saves the draft, and runs a typed review. knowledge_export_docx re-reviews the current Markdown and refuses export while any contract blocker remains.

Built-in contracts cover functional specifications, architecture documents, project briefs, onboarding guides, API references, ADRs, runbooks, test plans, incident reports, and release notes. custom remains available for a document with an explicit structure. Each contract defines purpose, default language and audience, required sections, minimum useful content, and type-specific checks; callers can override language and client-facing status without disabling structural validation.

The generated document is an output of agent memory, not its replacement. Confirmed facts belong in wiki/; source artifacts remain in docs/; delivery-ready Markdown and DOCX files belong in docs/deliverables/.

Optional OCR and semantic retrieval

Text, Markdown, JSON, YAML, CSV/TSV, XLSX, and PPTX normalization works locally. Images and PDFs require either an Ollama-compatible OCR service or a configured native OCR endpoint.

Common OCR variables:

Semantic retrieval is optional. Without it, deterministic lexical/graph/passage retrieval remains available. To enable an OpenAI-compatible embeddings endpoint, set all three required variables:

KNOWLEDGE_RAIL_EMBEDDING_BASE_URL=https://provider.example/v1
KNOWLEDGE_RAIL_EMBEDDING_MODEL=embedding-model
KNOWLEDGE_RAIL_EMBEDDING_DIMENSIONS=1536

Optional embedding variables are KNOWLEDGE_RAIL_EMBEDDING_API_KEY, KNOWLEDGE_RAIL_EMBEDDING_MODEL_VERSION, and KNOWLEDGE_RAIL_EMBEDDING_TIMEOUT_MS.

For Mermaid rendering, KNOWLEDGE_RAIL_MERMAID_NO_SANDBOX=true is intended only for controlled CI/container environments that cannot launch Chromium with its sandbox. Do not enable it by default on a workstation or shared host.

Compatibility

Modern product-level tools use the knowledge_* prefix. The wiki_* prefix is reserved for low-level operations that directly inspect or mutate canonical wiki/ pages. The legacy protocol adapter retains historical tool names for existing 2025-era clients without adding aliases to the modern catalog.

Development and verification

npm ci
npm run verify
npm run audit:runtime

Run all deterministic retrieval and quality gates:

npm run eval:retrieval:gate
npm run eval:hybrid:gate
npm run eval:widening:gate
npm run eval:source-coverage:gate
npm run eval:evidence-ir:gate
npm run eval:code-evidence:gate
npm run eval:recovery:gate
npm run eval:task-context:gate
npm run eval:semantic:gate
npm run eval:migration:gate
npm run eval:editorial:gate
npm run eval:documents:gate
npm run eval:tool-surface:gate

The benchmark fixtures and acceptance rules are documented in benchmarks/README.md. CI verifies Node.js 22 and 24, all regression gates, benchmark smoke tests, and the runtime dependency audit.

See CONTRIBUTING.md before opening a pull request and SECURITY.md for vulnerability reporting.

License

Licensed under the Apache License 2.0. You may use, modify, and distribute the project, including commercially, subject to the license terms and preservation of required notices. The license does not require derivative products to be open source.

Origins and acknowledgement

KnowledgeRail is an independent project. Its starting point was inspired in part by Andrej Karpathy's LLM Wiki idea file: an LLM maintains durable Markdown knowledge that compounds instead of reconstructing everything from raw sources on every query.

KnowledgeRail has since evolved into a distinct MCP 2.0 agent-memory system with bounded hybrid retrieval, coverage and gap reporting, Evidence IR, deterministic code evidence, migration support, and contract-driven document production. It is not affiliated with or endorsed by Andrej Karpathy. See ACKNOWLEDGEMENTS.md.