Using the CLI

The terminal dashboard, everyday commands, and example prompts for working with your coding agent.

The Dashboard

Run hamster with no arguments to open the terminal dashboard. It shows four tabs:

Dashboard — Brief and task counts, a progress bar for task completion, and recent activity across your project
Briefs — Browse briefs with a task list for each one. Press Enter to open a kanban view with tasks grouped by status
Blueprints — Read your architecture documents rendered as formatted markdown
Methods — Read your team conventions and processes

The dashboard watches for changes in real time. If a teammate updates a brief or moves a task in the browser, the dashboard reflects it within seconds.

Keyboard Shortcuts

Key	Action
`1` `2` `3` `4`	Jump to Dashboard / Briefs / Blueprints / Methods
`j` / `k` or arrows	Navigate lists
Enter	Open detail view or kanban
`e`	Open the selected file in your editor
Space	Cycle a task's status (pending, in progress, done)
`s`	Trigger a manual sync
`?`	Toggle help
`q`	Quit

Commands

Command	What it does
`hamster`	Open the terminal dashboard
`hamster sync`	Pull the latest project context once
`hamster sync --watch`	Keep local files updated continuously
`hamster sync --force`	Ignore cache and pull everything fresh
`hamster brief create "Title"`	Create a new brief
`hamster blueprint create "Title"`	Create a new blueprint
`hamster method create "Title"`	Create a new method
`hamster brief get`	View a brief's full content
`hamster task get HAM-123`	View a task's full details
`hamster team`	Switch between team accounts
`hamster status`	Show sync statistics, entity counts, and auth state
`hamster auth login`	Authenticate or switch accounts
`hamster auth logout`	Clear your local session
`hamster init`	Set up a new project (first-time only)

Example Agent Prompts

Once hamster sync --watch is running and your agent has the project context loaded, these kinds of prompts work naturally:

Starting work

"Read the brief for my-project and summarize what needs to be built"
"What's the next task I should work on?"
"Show me the acceptance criteria for HAM-042"

Building with context

"Implement HAM-015 following the patterns in the engineering-context blueprint"
"Work on the next task, using the methods in code-review-process"
"Check the brief requirements before starting HAM-023 — make sure nothing has changed"

Tracking progress

"Which tasks are still in progress?"
"How many tasks are done on the auth brief?"
"Mark HAM-015 as done and pick up the next one"

These prompts work because the agent reads the synced .hamster/ files directly. The brief requirements, task instructions, acceptance criteria, blueprints, and methods are all on disk as plain markdown — the agent does not need API access to answer these questions.

For agents that can also write back to Hamster (updating task status, creating subtasks), see MCP Server.

Tips