Tools

You don't pick tools. You describe the outcome you want, and Nova chooses the tools for the job. Every tool call shows up in the live work log under the answer, with the tool's name and a short preview of what it did, such as the search query, the command or the page address. Open a step to see its full result.

Which tools are available

• Nova can use every tool on this page by default.
• An agent can have its tools narrowed. Open the agent's editor, go to Tools, and turn tiles on or off. Tools with a lock are essential and always stay on. See Agents.
• A workflow step gets only the tools that step declares. See Workflows.
• Think is offered at High thinking effort and above. Understand and Execution Plan are offered at every level.
• MCP tools appear once you add an MCP server in Settings. See Integrations.

How tool calls look in the chat

Each step in the work log has a kind label: Agent, Tool, Code Run, Browser, Utility, Memory or Deep Reasoning. It also has a title, a tick when it is done, and a chevron to open it. Some tools draw a card of their own:

Code steps show their progress as they run, for example Running…, then Command finished.

Research and the web

For finding things out. Web search needs no key, and results come back ranked and citable, so answers can link to their sources.

• Web SearchSeveral searches at once across web, news, video, image and maps results, ranked and citable. No key needed.
• CrawlerCrawls many pages in a headless browser and pulls out their content.
• ScraperReads one or several pages or documents and answers questions about them.
• Knowledge SearchSearches the web or your own knowledge and memory.
• Knowledge DatabasesOne search across PubMed, arXiv, bioRxiv, medRxiv, ChEMBL, ClinicalTrials.gov, Open Targets, DailyMed, the NPI Registry and SEC filings.

Browser and desktop

For work that happens on a screen. The browser and the desktop both run on your instance, and you can watch them live in the side pane.

• Browser AutomationDrives the built-in browser you can watch: opens pages, clicks, types, reads and takes screenshots, in several tabs.
• Desktop ControlUses the cloud computer like a person: screenshot, mouse, keyboard, clipboard, opening apps.

Code and files

For running code and working with files. Code runs on your instance, and edits to text files show what changed.

• Code RunnerRuns shell, Python or Node.js on your instance, with terminals that stay open between steps.
• Coding AgentHands a coding job to a coding CLI you have connected (Codex, Claude Code, Gemini CLI, Cursor, Grok, OpenCode). Edits ask first.
• Text EditorReads, writes and patches text and Markdown files, and shows what changed.
• Office DocumentsCreates, edits, versions and exports Word, Excel and PowerPoint files (and their open formats).

Memory

For recalling what was said, decided or learned before. Saving to memory happens on its own after each reply; see Memory.

• Search MemorySearches the Journal, your knowledge and earlier analyses, and returns the evidence it found.

Agents and delegation

For handing work to specialists. You see each hand-off in the work log, and each agent's findings come back to Nova.

• Delegate TaskHands one task, or a batch in parallel, to specialist agents and brings back their findings.
• Spawn AgentsStarts up to 10 parallel workers and joins their results.
• Agent to AgentTalks to an outside agent that speaks the A2A protocol.
• Send MessageSends a message into another running chat.

Thinking and planning

For getting the job right before doing it. These are the steps that ask you questions or show you a plan to approve.

• ThinkExtra private reasoning before acting. Used at High thinking effort and above.
• UnderstandShows what it knows, what is missing and the questions that would change the route, and asks you.
• Execution PlanProposes a plan for your approval, then ticks off verified progress and replans when reality changes.

Operating the product

For changing things inside SyntrofAI itself, such as your goals, tasks, workflows, agents, teams, memory and screen.

• Platform ControlOperates goals, tasks, memory, workflows, sources, agents, teams and model settings for you. Changes ask for approval on a card.
• Widget ControlOpens, arranges and operates the apps on your screen: browser, office, terminal, maps, music and more.
• Create TaskCreates and updates scheduled tasks.
• SkillsFinds, loads, creates and installs Skills, the procedure packs that teach Nova a specialised job.

Media

For making images. Video generation lives in the AI Studio app; see Side-pane apps.

• Image GeneratorCreates or edits an image from a prompt with Gemini image models.

Social research

Read-only research on the main social platforms. Each one uses the credentials you add in Social → Settings; see Integrations.

• Reddit ResearchSearch, posts, comments, users, subreddits, media.
• X ResearchPosts, users, timelines, lists, followers, hashtags.
• Instagram ResearchProfiles, posts, stories, followers, hashtags, comments.
• TikTok ResearchUsers, videos, trending, hashtags, comments.
• YouTube ResearchVideos, channels, playlists, transcripts, comments.
• Telegram ResearchChannels, messages, participants, replies.
• Discord ResearchServers, channels, messages, members, threads.
• GitHub ResearchRepos, files, issues, commits, releases, trending.
• Social FeedSearches the posts your Social sources already collected.

Turn housekeeping

The small tools that start, pause and finish a turn.

• Final AnswerDelivers the reply and ends the turn.
• User NotificationShows you a short notice without ending the turn.
• Read ResultRe-reads an earlier result instead of running the step again.
• PauseWaits a set time or until something is ready.
• MCP toolsAny tool from an MCP server you add in Settings → MCP/A2A.

Approvals and safety

Tools that change things ask before they act. What happens depends on the tool.

Related