MCP tool catalog

The MCP server exposes tools across projects, tasks, code analysis, knowledge, databases, and memory. This page catalogs the ones you will use most; database tools get their own page.

Unless noted, every tool takes a project_id (defaulting to the current project where the server supports one) and returns JSON.

Session

Ontology / faceted knowledge

These map one-to-one onto the Knowledge API.

ontology_discover(project_id)
  -> { project_id, repos, sources, providers }

ontology_search(project_id, query, sources?, providers?, repo_id?, since?, top_k=10)
  -> { hits: [{ source, provider, ref, title, snippet, score, url, timestamp, resource_uri }], count, mode }

ontology_fetch(project_id, ref)
  -> { ref, source, title, content, url, metadata }

ontology_activity(project_id, since?, sources?, providers?, top_k=20)
  -> time-ordered events (newest first) from work_items, commits, prs

Projects & workspaces

list_projects(workspace_id?), get_project_info(project_id?), list_project_members(project_id?), list_workspaces(), get_workspace_info(workspace_id?), compare_projects(project_ids[]), get_active_llm(project_id).

Tasks, epics & project context

list_tasks(project_id, epicId?, status?, assignedTo?), show_task_board(project_id, ...), get_task_details(project_id, task_id), list_epics(project_id, status?), get_next_task(project_id, ...), get_project_context(project_id), graph_search(project_id, query), graph_query(project_id, node_name).

Code, documents & memory

search_code(project_id, analysis_id, query, top_k=10, ...), search_documentation(...), list_documents(project_id), search_uploaded_documents(project_id, query, top_k=10), get_document_content(project_id, storage_path).

memory_search(project_id, query, limit=10, types?)
  -> { items: [{ id, memory_type, what, why, where, learned, status, created_at }], count, mode }
memory_stats(project_id)
  -> { total, by_type: [{ memory_type, count }] }

Note the two search families: graph_search and graph_query traverse the Manager entity graph (tasks, epics, modules the manager created), while ontology_search fans across the Ontology Layer sources. They answer different questions.

Next: Database tools.