WikiStrata
Ingests Confluence page trees and attachments as source data, builds structured Markdown wikis, and exposes them via MCP.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@WikiStrata search for refund policy"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
WikiStrata
Build a Markdown wiki and MCP layer from a Confluence tree.
WikiStrata is a public-facing, Confluence-first toolkit for turning internal documentation into structured Markdown, searchable indexes, and agent-friendly MCP context.
Language versions:
This is a prototype and a concept. Take it, adapt it, make it yours.
WikiStrata is currently Confluence-first.
It is built to help a team connect a Confluence tree, ingest it into immutable source files, rebuild it into Markdown, and expose that wiki through MCP.
Goals
ingest Confluence page trees as immutable Layer 1 source data
rebuild them into a structured Layer 2 wiki using schema-driven rules
expose the wiki through stdio and HTTP MCP transports
provide reusable onboarding via
init-style schema suggestion workflows
Related MCP server: personal-llm-wiki-kit
Current public-safe scope
generic Confluence ingest and attachment extraction
schema-driven sample build pipeline
Whoosh-backed search indexing
public CLI for init, ingest, build, index, stats, and MCP entrypoints
stdio MCP server
HTTP MCP server with CIDR and token gates
synthetic sample schema and test coverage
a tiny committed sample vault for GitHub browsing
Screenshots
CLI

Generated Summary

Obsidian Graph View

Quickstart
cd <project-root>
python -m venv .venv
.venv\Scripts\activate
python -m pip install -e .Then:
wikistrata init --write-starter-schema
wikistrata sample-project .\wikistrata-sample
cd .\wikistrata-sample
wikistrata build-wiki
wikistrata build-index
wikistrata search "refund policy"
wikistrata rebuild
wikistrata-mcpDocumentation
docs/CONFLUENCE_FIRST.mddocs/QUICKSTART.mddocs/SCHEMA_GUIDE.mddocs/SAMPLE_DATASET.mddocs/SCREENSHOT_GUIDE.mddocs/INIT_ONBOARDING_DESIGN.mddocs/REPO_BOUNDARY.mddocs/RELEASE_CHECKLIST.mddocs/PUBLIC_CONVERSION_AUDIT.md
Included samples
examples/sample-vault/a very small committed markdown sample for GitHub browsing
wikistrata sample-project ...the fuller synthetic demo flow for screenshots, Obsidian, and CLI demos
What is happening in this workspace
This workspace is being built with a whitelist strategy. Only generic engine code, generic schema examples, synthetic sample data, and public documentation should land here.
See:
docs/PUBLIC_CONVERSION_AUDIT.mddocs/WHITELIST.mddocs/MIGRATION_PLAN.md
Release gates
The project should not be published until all of the following are true.
no company names, brand names, internal URLs, page IDs, or internal examples remain
schema files are synthetic and domain-neutral
tests use synthetic fixtures only
documentation is rewritten for public users
the package name, CLI help text, and MCP server metadata are public-ready
Immediate priorities
Move only generic core modules into
src/wikistrata.Replace internal schema and fixtures with synthetic examples.
Rewrite public docs around setup, schema customization, and MCP usage.
Rename the package and CLI surface from internal names to public names.
Deferred work
These items do not block the split itself.
build_wiki.pyrefactorproduction LLM synthesis automation
PPT image extraction
auth modernization beyond Basic Auth
Project Status
WikiStrata is an early public release.
The public surface is intentionally narrow and strong:
Confluence-first ingest
Markdown wiki generation
Whoosh-based search
stdio and HTTP MCP access
Broader multi-source onboarding and richer automation can come later.
Available Tools
10 toolsbrowse_treeC
Browse the wiki tree by node id.
Inputs: node_id, max_depth.
get_tree_pathC
Resolve the tree path for a document by path, title, or source page ID.
Inputs: path, title, page_id.
inspect_conceptC
Inspect a concept from the wiki concept registry.
Inputs: name (required).
inspect_entityC
Inspect an entity from the wiki entity registry.
Inputs: name (required).
read_documentB
Read a wiki document by path, title, or source page ID.
Inputs: path, title, page_id.
reload_wikiC
Reload wiki markdown and registry files from disk.
Inputs: no named parameters.
search_in_subtreeC
Search only inside a selected tree node.
Inputs: ancestor_id (required), query (required), doc_type, year, +2 more.
search_wikiB
Search wiki documents by title, path, tags, and body text.
Inputs: query (required), doc_type, year, solution, +1 more.
wiki_statsA
Return document and registry counts for the current wiki.
Inputs: no named parameters.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 10 tools
Each tool targets a distinct aspect of the wiki: tree navigation, path resolution, concept/entity inspection, document reading, related documents, search (general and subtree), reloading, and statistics. No overlapping purposes.
All tool names follow a consistent verb_noun pattern in snake_case (e.g., browse_tree, inspect_concept, search_wiki). Even compound names like get_tree_path and search_in_subtree maintain the pattern.
10 tools is well-scoped for a wiki server, covering browsing, searching, reading, inspection, and admin functions without being excessive or insufficient.
The tool set covers reading, searching, and browsing, but lacks any create, update, or delete operations for documents or registry entries. This is a notable gap for a wiki system, though the server may be intended as read-only.
Related MCP Connectors
Serve a folder of Markdown notes as an MCP server: hybrid search, reading, and sourced answers.
MCP server for querying Forkast documentation
Hosted markdown project wikis your team's AI assistants read, search, and update over MCP.
Read-only MCP server for the OrchestKit docs: full-text search + Markdown fetch. No auth.
Related MCP Servers
- FlicenseNot gradedqualityBmaintenanceMCP server for searching, reading, listing, writing, and appending Markdown wiki content using RAG with ChromaDB.-
- FlicenseNot gradedqualityDmaintenanceEnables querying a local wiki through MCP-style tools, including search, Q\&A, and knowledge map, with a web viewer for human browsing.-
- AlicenseNot gradedqualityDmaintenanceMCP server for searching and retrieving pages from Atlassian Confluence.30 npm1MIT
- FlicenseAqualityDmaintenanceLightweight MCP server that crawls Confluence subtrees and serves cached Markdown, enabling search and retrieval of Confluence pages via MCP tools.4-