This walkthrough takes you from nothing installed to a productive Synsci CLI session in under five minutes. You’ll run a single-command install, authenticate to your dashboard, pick a research mode, and open the web UI.
1

Install in one command

Run the all-in-one installer:
npx synsci
This downloads the binary, walks you through dashboard authentication, and pre-fetches your synced credentials. When it finishes, synsc is on your PATH and ready.See Installation for npm, Homebrew, or curl alternatives.
2

Verify the connection

Confirm authentication:
synsc connect status
Healthy output prints your account email, current credit balance, and the synced credentials available. Hugging Face, Weights & Biases, Modal, Lambda, AWS, and GCP all show up if you have them set in the dashboard.
Add or rotate credentials at cli.syntheticsciences.ai. The CLI picks them up on the next session start.
3

Pick a research mode

Synsci CLI works best when you tell it what kind of work you’re doing. Pick a research mode before starting a session:
synsc web --mode research     # general scientific research
synsc web --mode physics      # simulation, PDEs, dynamical systems
synsc web --mode chemistry    # drug discovery, molecular design, docking
synsc web --mode biology      # bioinformatics, structure prediction, omics
synsc web --mode flywheel     # build a specialized LLM from production data
Setting a mode primes the agent with domain conventions, activates the right bundled tools, and routes calls to the backend best suited for that work. You don’t need to think about underlying providers.
If you don’t know which mode you want yet, run synsc web without --mode for the general-purpose agent. You can switch modes mid-session with /mode <name>.
4

Send your first prompt

synsc web opens the browser UI on http://localhost:4444. Type a prompt and press Enter:
Read README.md and summarize what this repo does in three bullets.
The agent reads the file, streams its response, and prompts you for the next message. You’re in the build agent.
5

Switch agents

Two top-level agents are available:
  • build has full filesystem and shell access (default).
  • plan is read-only for analysis without mutations.
Switch with the agent toggle in the UI header, or press Tab in the TUI.Mention @general in a message to delegate a search or multi-step task to a sub-agent that runs in its own context window:
@general find every occurrence of LearningRateScheduler in this repo
6

Inspect usage

After running a few prompts, check what you spent:
synsc stats
stats aggregates credit usage per session and across time windows. The full audit trail lives on the dashboard.
7

Run one-shot prompts (optional)

For scripted workflows, skip the UI and pipe a single prompt with synsc run:
synsc run --mode ml "Refactor utils/log.ts to use structured logging"
The agent runs the prompt to completion and exits. Useful for git hooks, CI, and shell pipelines.

​
What’s next