COMMANDS
Every capability, one context lake.
Run fulcra --help for the current command list. The CLI covers the whole loop: ask what changed since your last run — including what your other agents wrote — read it, record new data, manage files and tags, and create or revoke shares. Output is JSON by default, for piping into jq — or straight into an agent.
What changed since the last loop
One ask summarises everything that changed across a window you name — 3h, 2 days, 1 week, or two timestamps. It returns which kinds of records arrived, how many, and which files moved — including what your other agents recorded. This is how an agent starts where it left off instead of starting over.
Read anything in the lake
The catalog says what exists and describes each kind of record. Read raw samples or a calculated time series over any window. Derived views — calendars, locations, summaries — are queries over the same catalog, not separate products.
Write back
Record to any recordable type — a single value, or structured data piped in. Delete what shouldn't stay. Organise with tags, and manage files end to end: upload, download, restore. What an agent records lands where you and your other agents can see it.
Share it, and revoke it
Share selected data types — or a single file — with other Fulcra users, see what's shared in and out, and revoke any of it. Grant access, revoke it, change agents — a two-way door, from a shell.