Outline Wiki MCP Server

Instructions

Implementation Reference

• src/lib/handlers/smart.ts:22-83 (handler)

  The core handler function that implements the sync_knowledge tool logic: fetches documents from Outline API, retrieves full text content, and synchronizes them to the vector store (brain) for RAG capabilities.

  async sync_knowledge(args: { collectionId?: string }) {
    if (!brain.isEnabled()) {
      return { error: ERROR_MESSAGES.SMART_FEATURES_DISABLED };
    }
  
    // Step 1: Fetch document list from Outline
    const payload: Record<string, unknown> = { limit: 100 };
    if (args.collectionId) {
      payload.collectionId = args.collectionId;
    }
  
    const { data: docList } = await apiCall(() =>
      apiClient.post<OutlineDocument[]>('/documents.list', payload)
    );
  
    if (!docList || docList.length === 0) {
      return { message: ERROR_MESSAGES.NO_DOCUMENTS_FOUND, synced: 0 };
    }
  
    // Step 2: Fetch full content for each document (list API may truncate text)
    const wikiDocs: WikiDocument[] = [];
    let fetchErrors = 0;
  
    for (const doc of docList) {
      try {
        const { data: fullDoc } = await apiCall(() =>
          apiClient.post<OutlineDocument>('/documents.info', { id: doc.id })
        );
  
        if (fullDoc && fullDoc.text) {
          wikiDocs.push({
            id: fullDoc.id,
            title: fullDoc.title,
            text: fullDoc.text,
            url: `${baseUrl}${fullDoc.url}`,
            collectionId: fullDoc.collectionId,
          });
        }
      } catch {
        fetchErrors++;
      }
    }
  
    if (wikiDocs.length === 0) {
      return {
        message: ERROR_MESSAGES.NO_DOCUMENTS_WITH_CONTENT,
        synced: 0,
        errors: fetchErrors,
      };
    }
  
    // Step 3: Sync to brain (vectorize)
    const result = await brain.syncDocuments(wikiDocs);
  
    return {
      message: `Successfully synced ${result.documents} documents (${result.chunks} chunks).`,
      documents: result.documents,
      chunks: result.chunks,
      skipped: docList.length - wikiDocs.length,
      errors: fetchErrors,
    };
  },

• src/lib/schemas.ts:142-144 (schema)

  Zod schema definition for the sync_knowledge tool input, which accepts an optional collectionId.

  export const syncKnowledgeSchema = z.object({
    collectionId: collectionId.optional(),
  });

• src/lib/tools.ts:200-204 (registration)

  Registers the sync_knowledge tool in the allTools array, providing name, description, and schema reference for MCP tool definition.

  createTool(
    'sync_knowledge',
    'Sync documents to vector store for AI-powered search. Run this before using ask_wiki.',
    'sync_knowledge'
  ),

• src/lib/schemas.ts:200-200 (schema)

  TypeScript type derived from the syncKnowledgeSchema for input validation.

  export type SyncKnowledgeInput = z.infer<typeof syncKnowledgeSchema>;

• src/lib/schemas.ts:244-244 (schema)

  Includes the sync_knowledge schema in the central toolSchemas map used by tool definitions.

  sync_knowledge: syncKnowledgeSchema,