Documentation
MCP Tools Reference
49 tools across 9 categories. Every tool is available in the free tier — you can call them yourself, or just ask Cerebro in chat and it picks the right ones for you.
Memory
6 tools — Persistent memory & recall
searchHybrid semantic + keyword search across every memory, with recency boostingsave_conversation_ultimatePersist a conversation with full extraction of facts, entities, and decisionssearch_knowledge_baseQuery the central knowledge base of distilled facts and discoveriesget_entity_infoEverything known about a specific entity — a person, tool, server, or projectget_timelineChronological timeline of actions and decisions for any time periodfind_file_pathsRecover every file path mentioned across conversations, with its contextAgents
7 tools — Deploy & coordinate agents
spawn_agentDeploy a specialized agent — worker, researcher, coder, analyst, browser, or SpecOpsagent_statusLive status, progress, and streamed output for any running agentagent_groupsOrganize agents into projects and missions you can track over timeorchestrateCoordinate multi-agent teams that share one workspace via claim/releaselistenersStanding agents that watch for a trigger and act the moment it firesrelay_messageMessage a running agent, or relay context between chats and agentsstop_agentPause, resume, or terminate any agent on demandBrowser
5 tools — Web automation & scraping
session_createLaunch an isolated Chrome session — visible on a monitor or fully headlessnavigateDrive a page to any URL and wait for it to settleclick_typeInteract with elements by selector — click buttons, fill forms, press keysextractDump the DOM or run JavaScript to scrape exactly what you needscreenshotCapture full-page or element screenshots, saved to disk and returned inlineGoals
4 tools — Goals & project tracking
goalsDetect, track, and proactively surface context for your active goalsprojectsTrack projects, their members, and status across long horizonsproject_evolutionVersion history and milestones for any project over timeget_suggestionsProactive next-step suggestions drawn from current context and past patternsAutomations
4 tools — Scheduled tasks
scheduleCreate scheduled tasks — cron, interval, or one-time fire timesrun_automationTrigger any automation on demand, outside its scheduleautomation_statusRun history, last result, and next fire time for every automationtriggersEvent-driven automations that fire when a condition is metDevices
4 tools — Remote machine control
deviceRegister and manage the machines on your device networkssh_execRun commands on a remote registered machine over SSHsearch_by_deviceFilter memories by the device they originated ondevice_statusHealth and connectivity for each machine in the networkCapability Packs
5 tools — Extensible abilities
list_packsBrowse the capability packs available to extend Cerebroactivate_packEnable a pack for a chat or session to unlock its abilitiescreate_packAuthor a brand-new pack to teach Cerebro skills it didn't havepack_toolsInspect the endpoints and instructions a pack addsmcp_managementRegister and hot-swap MCP servers per chatAnalysis
7 tools — Reasoning & analysis
reasonActive reasoning over memories — generate insights without promptingcausalBuild and query causal models to trace why things happenanalyzeDetect patterns, knowledge gaps, and skill-development trendsconsolidateCluster episodes, create abstractions, and strengthen memory linksself_modelContinuous self-awareness — track confidence, uncertainty, and limitsmeta_learnOptimize retrieval by learning which search strategies actually workpredictPredict outcomes using causal models and historical patternsSystem
7 tools — Platform management
system_health_checkFull diagnostic of every memory-system component at oncerebuild_vector_indexRebuild the FAISS semantic index for optimal retrievalqualityFind duplicates, compute health scores, and tidy the memory storeconfidenceFact confidence scoring, provenance tracking, and contradiction detectionprivacyScan for secrets, manage redaction, and audit sensitive datadecayArchive stale memories while protecting the ones that matterself_reportGenerate self-improvement reports and track performance metricsNote: The memory MCP server is open source — drop it into your own Claude Code for persistent memory. Pro adds the desktop app, the visual memory explorer, unlimited agents, and full autonomy — but the tools themselves are always yours.