AI knowledge base for JetBrains AI.
Bring your team’s wiki into IntelliJ IDEA, PyCharm, WebStorm and every other JetBrains IDE. Connect the Perfect Wiki MCP server to AI Assistant or Junie, and your coding agent works from the same specs and standards your team does.
Connect JetBrains AI in two minutes.
Junie reads MCP servers from mcp.json and supports OAuth sign-in. For AI Assistant, paste the server as JSON in its MCP settings together with an API token header.
- →Junie: OAuth sign-in, config per user or per project
- →AI Assistant: add the server as JSON in Settings
- →One setup across IntelliJ IDEA, PyCharm, WebStorm, GoLand and more
Junie builds from the spec — the one in your wiki.
Product specs, domain rules and coding standards stop living in a separate tab. The agent fetches the page it needs, implements against it, and can note what changed.
- →Business rules read from the source of truth, not a pasted excerpt
- →Coding standards applied consistently across the team
- →Specs and changelog pages updated when you allow writes
PricingService.kt with a test for each rule, including the no-stacking case.How to connect Perfect Wiki to JetBrains AI.
- 1In Junie, open Settings | Tools | Junie | MCP Settings and click + — or edit
~/.junie/mcp/mcp.json(all projects) or.junie/mcp/mcp.json(one project). - 2Add
perfect-wikiwith"url": "https://api.perfectwiki.xyz/mcp". To authorize from the Junie CLI, run/mcp, select perfect-wiki and choose Authorize, then sign in to Perfect Wiki and allow access. - 3For AI Assistant, open Settings | Tools | AI Assistant | Model Context Protocol (MCP), click Add and paste the JSON with your API token.
- 4Ask in chat — for example “Check Perfect Wiki for our Kotlin error-handling guidelines.”
More details in the Perfect Wiki MCP server guide and JetBrains AI's MCP documentation.
Generate a token in the Perfect Wiki settings of the knowledge base you want to expose. Each token is scoped to a single knowledge base and has read & write access — sign in with OAuth instead to cover several knowledge bases with one connection.
What your team does with Perfect Wiki in JetBrains AI.
Business rules straight from specs
Invoicing, pricing and eligibility logic is implemented from the rules page product maintains — and checked against it later.
Legacy code, explained
Pair the code with the architecture overview and module ownership docs so the agent can explain why things are built the way they are.
Standards in every review
Ask the agent to review changes against your language style guides and conventions before a human reviewer spends time on them.
Ask a question in plain language and get an AI answer from your SOPs and FAQs, with the source pages.
See which knowledge bases the connection covers and the access level for each.
List every page in a knowledge base.
Read the full content of a specific page.
Create a new page — draft SOPs straight from chat.
Update an existing page, so SOPs never go stale.
Remove pages that are no longer needed.
Get the name and access level of one knowledge base.
Start from a ready-made SOP. Keep it alive here.
Download any of our 294 standard operating procedure templates in Word, PDF, Excel or CSV, no sign-up needed. Then put it in Perfect Wiki so your team can find it and ask about it.
Browse all 294 SOP templates →Frequently asked questions
Yes, both connect to remote MCP servers. Add it in Settings | Tools | AI Assistant | Model Context Protocol (MCP), or in Settings | Tools | Junie | MCP Settings — Junie also reads ~/.junie/mcp/mcp.json.
Junie documents OAuth sign-in for remote servers, so you can connect without a token. AI Assistant doesn’t document OAuth for MCP servers, so the reliable option there is a Perfect Wiki API token in the Authorization header.
Any JetBrains IDE where AI Assistant or Junie is available — including IntelliJ IDEA, PyCharm, WebStorm, GoLand, PhpStorm and Rider. The MCP configuration is the same everywhere.
Uploaded files are a snapshot: they go stale the moment a policy changes, and they stay in one chat or project. Through the MCP server, JetBrains AI reads the current version of your SOPs and FAQs from Perfect Wiki every time, links to the source page, and only reaches the knowledge bases you connect.
Only if you allow it. When you sign in with OAuth you choose Read only or Read & write on the consent screen, and writes only work in knowledge bases where you can edit. Legacy API tokens always have full access to their one knowledge base. Most MCP clients also ask for your confirmation before running a tool that writes.
Yes, with a single connection. When you sign in, choose all knowledge bases you can access (including ones you get access to later) or only the ones you select. JetBrains AI can search across all of them at once and has to name the target knowledge base for every change. API tokens cover one knowledge base each, so add one server entry per token.
The Perfect Wiki MCP server only forwards the requests JetBrains AI makes. Tool results — the pages and answers it fetches — are passed to the model as part of the conversation and handled under your AI provider's data policy.
The MCP server is included in every Perfect Wiki plan, and the same API rate limits apply. You can try it with the 14-day free trial — no credit card required.
Didn't find what you're looking for? Contact our support →
Put your SOPs to work in JetBrains AI.
Keep your SOPs, manuals and FAQs in Perfect Wiki — and give your whole team answers right where they work.