FusionAI CLI brings the platform to your machine. Work on local files and repos from the terminal while memory, coordination, triggers, and the broader agent platform remain anchored in the cloud. It is conceptually similar to local agent workflows, but it is FusionAI's own implementation — not a wrapper for other agent runtimes.
FusionAI CLI is not limited to one environment. Start inside your repo on your machine, then hand work off to your cloud workspace for long-running agents, scheduled automations, webhooks, and team operations — without losing memory, context, or skills.
Run agents directly in your local repo, inspect diffs, execute tests, debug failures, and iterate with full control over your environment.
Move from local execution to cloud agents when the job should keep running after you close your laptop — follow-ups, monitors, webhooks, and scheduled workflows.
Same agents. Same skills. Same memory. Same workspace context. Start locally, continue in the cloud, and pick the execution environment that fits the job.
Start on your machine, continue in the cloud, or wire both together. FusionAI CLI gives you one interface across terminal sessions, local repos, CI pipelines, and live event-driven automations.
Point an agent at a task and let it run. The CLI streams every step — reasoning, tool calls, and results — straight into your shell.
Drop the CLI into any repo and ask an agent to map the architecture, trace a flow, or explain a module. It reads, indexes, and answers in plain language — with file references.
Describe what you want — a fix, a script, a refactor — and the agent writes, runs, and tests it on a real machine. You review a diff or a PR, not a wall of suggestions.
Schedule agents or wire them to events. Stale PRs, failing builds, overdue tickets — the CLI lets agents own the loop and keep working after you log off.
Use local execution for hands-on work. Use cloud execution for persistence, schedules, webhooks, and team workflows. FusionAI lets you move between both without rebuilding your stack.
Execute agents inside your local repo with full control. Read your real files, run tests, inspect diffs, and debug failures on your own machine — no upload, your code never has to leave your laptop.
Delegate long-running work to your cloud workspace. Agents keep running 24/7, fire from schedules and webhooks, and respond to incoming events long after you close your laptop.
Agents, skills, memory, and workspace context sync both ways. Start a task locally and finish it in the cloud — or vice versa. The same context follows you across both environments.
A single curl command installs the CLI with zero dependencies — no Docker, no Node, no config. Up and running in seconds on macOS and Linux.
Run fusionai init once and your local workspace mirrors the cloud — agents, skills, and memory available offline, reconciled when you reconnect.
Headless, deterministic runs with --ci: scoped tokens, structured JSON output, and exit codes that fail your pipeline when an agent does.
Run curl -fsSL https://fusionai.now/install.sh | bash on macOS or Linux. The installer drops a single binary at /usr/local/bin/fusionai and creates a config directory at ~/.fusionai/. No Docker, Node, or other dependencies required. Then run fusionai init to connect your workspace.
macOS, Linux, and Windows are supported. On Windows, the native PowerShell installer at https://fusionai.now/install.ps1 is the primary path. WSL2 remains available as a fallback if your team standardizes on Linux shells.
Yes. Agents run inside a sandboxed local runtime and only touch the files and environment you grant them. Credentials are stored encrypted in ~/.fusionai/, scoped per workspace, and never written to your shell history. You can review every tool call before it executes with --confirm.
The CLI inherits FusionAI's model-agnostic execution. You don't have to pick a provider for every run — FusionAI chooses the right model or tool mix for the task while keeping your agent, skills, and memory intact across local and cloud work.
Absolutely. Drop it into GitHub Actions, GitLab CI, or any shell-based pipeline. Use --ci for headless, deterministic runs with structured JSON output and meaningful exit codes, and authenticate with a scoped FUSION_KEY environment variable instead of an interactive login.
Agents read your local repository, files, and environment directly — no upload needed. Long-term and short-term memory sync with your cloud workspace, so an agent remembers prior runs whether you start them locally or in the cloud. You control exactly which paths an agent can access.
The CLI itself is free and included with every FusionAI account, including the free tier. You only pay for agent usage on your plan — local runs draw from the same workspace quota as cloud runs. See pricing for details.
Install once and run agents, workflows, and automations from anywhere you work. Free with every FusionAI account.
powershell -NoProfile -ExecutionPolicy Bypass -Command "iwr https://fusionai.now/install.ps1 -UseBasicParsing | iex"