The command line
Everything Tharas can do is reachable from a terminal, and the read-only commands cost nothing, so exploring is free.
The shape of the command set#
- Projects
- Add, list, rename, hide and remove the repositories Tharas knows about.
- Understanding
- Run the curator over a repository, accept its read, build the memory index, and ask what the planner can see.
- Work
- A direct build for something small, or a goal for anything non-trivial — which drafts a specification and waits.
- Decisions
- List what is pending, approve or reject it, and advance a plan. Advancing is idempotent, so re-running is safe.
- Programs
- For a large objective spanning many plans: the live graph of units, and controls to pause, resume or cancel.
- Measurement
- The resolution scorecard and the logs behind it.
A first session#
$ tharas projects add myproject /path/to/repo $ tharas index $ tharas goal "Add CSV export to reports" --budget 90m gate 0 specification drafted $ tharas proposals $ tharas approve <id>
The read-only commands are free#
Status, proposals, the project map and the scorecard make no model calls. They are worth running before anything that spends, particularly the map — a plan built on an incomplete picture produces work in the wrong place.
Common questions
Is the CLI the full product?
Yes. The dashboard and Telegram are different views of the same engine, not a subset or a superset.