GitScrum Knowledge MCP. Search notes, wikis, and chats with one query.

GitScrum Knowledge connects your AI agent directly to every piece of knowledge in your project workspace. You don't just talk to a chat app or pull data from a ticketing system; you give your agent access to the full context—the decisions, the documentation, and the discussions that built this thing.

It lets your agent search notes, wikis, tasks, and conversations all at once.

Global Context Retrieval
When you need an answer, the global_search tool queries everything simultaneously: notes, wiki pages, tasks, and discussion threads. You run one command and get a single result set that aggregates every relevant finding, meaning your agent never has to switch context or guess where the information lives.

Structured Knowledge (Notes)
Your team's decisions shouldn't live in chat logs; they belong in persistent memory. The create_note tool lets you write down temporary thoughts or permanent architectural decisions, treating them like structured facts for your agent to reference later. You can refine this memory over time by using update_note, which appends new context without erasing old data.

If a decision is finalized, the agent helps manage its visibility; toggle_note_share moves a note from private agent memory out into public team knowledge.

You'll keep your notes organized with dedicated folders. Use create_note_folder to set up groupings (like 'ADRs'), then use list_note_folders to see what you've got, and move_note_to_folder or rename_note_folder to keep things tidy. If a note needs re-siting, update_note helps, but if it's completely obsolete, delete_note permanently clears the record.

To track how ideas change, you can't just read the final version. The note_revisions tool retrieves the full history of changes made to a specific note, letting your agent show exactly how an idea evolved from concept to execution.

Formal Documentation (Wikis)
For company standards and manuals—the stuff that supports nested content and markdown formatting—you build out formal documentation using wiki pages. You initiate this process with create_wiki_page, building robust processes far beyond a simple memo. These aren't meeting notes; these are the authoritative guides. The agent can fetch the complete, detailed text of any document using get_wiki_page.

If someone messes up and corrupts the documentation, you don't sweat it; wiki_revisions gets the full history, and you use restore_wiki_revision to revert the page back to a clean version.

You can edit existing guides with update_wiki_page, modifying content without creating redundant copies. If a document is totally wrong and needs removal, delete_wiki_page handles it entirely.

Team Communication Context (Channels & Discussions)
Your agent manages the flow of team conversation so you don't have to babysit Slack threads. You can list all available discussion channels using list_channels, or create a dedicated space with create_channel. The toolset keeps conversations linear and searchable: use send_message to post an immediate update, or reply directly within the flow of conversation using reply_to_message.

If you need to know what was said in that channel last month, search_channel_messages digs into the content. You can also retrieve all messages contained within a specific thread using channel_messages, and get all replies made in response to one message via thread_replies.

For general team visibility, the agent keeps track of all ongoing or completed project discussions by listing them with list_discussions. To review what was said in any given channel, you use get_channel to pull high-level details. You can also search for specific keywords within a single discussion channel using search_channel_messages, or run a focused keyword deep dive across all stored wiki content using search_wiki.

Putting It All Together
When you ask your agent, 'What were the payment gateway decisions and what did we tell marketing about it last month?', it doesn't guess. It runs global_search. That query pulls data from notes (decisions), wikis (documentation standards), channels (search_channel_messages), and discussions—and gives you one answer that connects all those dots.

You don't just get a list of documents; you get the synthesized truth about your project.