KnowledgeRail
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| WIKI_ROOT | No | Root directory of the project whose knowledge is managed. Omit to use the server process working directory. | |
| KNOWLEDGE_RAIL_OCR_MODE | No | OCR mode: 'ollama' (default) or 'native'. | ollama |
| KNOWLEDGE_RAIL_OCR_MODEL | No | OCR model to use. | glm-ocr:latest |
| KNOWLEDGE_RAIL_NATIVE_HOST | No | Native OCR base URL. | http://localhost:5002 |
| KNOWLEDGE_RAIL_OCR_RETRIES | No | Retry count for OCR. | |
| KNOWLEDGE_RAIL_OLLAMA_HOST | No | Ollama base URL for OCR. | http://localhost:11434 |
| KNOWLEDGE_RAIL_OCR_TIMEOUT_MS | No | Positive request timeout in milliseconds for OCR. | |
| KNOWLEDGE_RAIL_EMBEDDING_MODEL | No | Embedding model name. Required to enable semantic retrieval. | |
| KNOWLEDGE_RAIL_EMBEDDING_API_KEY | No | Optional API key for the embeddings endpoint. | |
| KNOWLEDGE_RAIL_EMBEDDING_BASE_URL | No | Base URL for OpenAI-compatible embeddings endpoint. Required to enable semantic retrieval. | |
| KNOWLEDGE_RAIL_MERMAID_NO_SANDBOX | No | Set to 'true' to disable Chromium sandbox for Mermaid rendering. Only for controlled CI/container environments. | |
| KNOWLEDGE_RAIL_EMBEDDING_DIMENSIONS | No | Dimensions of the embedding model. Required to enable semantic retrieval. | |
| KNOWLEDGE_RAIL_EMBEDDING_TIMEOUT_MS | No | Optional timeout in milliseconds for embeddings requests. | |
| KNOWLEDGE_RAIL_EMBEDDING_MODEL_VERSION | No | Optional model version for the embeddings endpoint. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| knowledge_contextC | Evidence/gaps, pages, search, and graph. |
| knowledge_pageC | Page CRUD and durable log. |
| knowledge_filesA | Source files/PDFs: list, read, normalize to Markdown. |
| knowledge_ingestD | Source ingestion, claims, coverage, recovery. |
| knowledge_codeD | Code index, symbols, callers, fallback. |
| knowledge_document_contextD | Document plans and section evidence. |
| knowledge_documentC | Evidence-backed document writing/review. |
| knowledge_adminD | Workspace setup, maintenance and diagnostics. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| plan_document | Generate an editorial plan (sections, writers, context packs, checklist) for a document in docs/deliverables. |
| plan_dev_report | Return the required development-report template to complete before wiki ingestion. |
| prepare_knowledge_update | Generate a valid wiki draft from a gap and the available evidence. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| wiki-schema | Wiki conventions and workflow |
| wiki-log | Wiki operations log |
TDQS
Scored across 8 tools
Most tools have clearly distinct purposes, but knowledge_context and knowledge_document_context have some overlap in terms of 'context', though descriptions indicate different aspects (evidence/gaps vs. document plans). Overall, an agent can reliably distinguish them.
All tools follow a consistent 'knowledge_<noun>' pattern with straightforward nouns (code, admin, context, document, files, ingest, page). This is highly predictable and clean.
8 tools is well within the ideal range for a knowledge management server, covering ingestion, coding, document writing, context, and administration without redundancy or bloat.
The tool surface covers core workflows: ingestion, file handling, document writing/review, page CRUD, and context/graph search. Minor gaps might exist (e.g., explicit document deletion or search-only tool), but most operations are covered.