Lemma CLI (lemma-terminal)

lemma is the command-line and terminal-UI for Lemma — build and operate pods (a team's tables, files, functions, agents, workflows, schedules, connectors, surfaces, and apps) from your terminal, and drop Lemma's agent skills straight into your coding agent so it can build pods for you.

Install

uv tool install lemma-terminal

This installs the lemma command globally. (Don't have uv? See astral.sh/uv.)

lemma --version

Quickstart

lemma auth login            # authenticate against the default cloud server
lemma orgs select --save-default   # pick your organization
lemma pods select --save-default   # pick the pod to work in
lemma describe              # inventory the selected pod

From there, the command surface mirrors the resource model — lemma <resource> <verb>:

lemma tables list
lemma files ls /knowledge
lemma agent chat            # talk to the pod's default agent
lemma pod init my-pod       # scaffold a new pod bundle on disk to import

Working across several pods at once (e.g. a coding agent per repo)? lemma app init and lemma pods create --with-starter write committed .lemma.<server>.env files that bind the folder per server, so every command from that tree targets the right pod — no per-shell exports, no global switching. The same repo drives local and cloud:

lemma pods describe                    # local pod  (from .lemma.local.env)
lemma --server lemma-cloud apps deploy # cloud pod  (from .lemma.lemma-cloud.env)
lemma config show                      # shows the resolved server, pod, and files

Add --json to any command for machine-readable output, and --full to expand folded fields. See SETUP.md for cloud/local server configuration, environment variables, project .lemma.<server>.env files, and the Textual TUI (lemma tui).

Install Lemma skills into your coding agent

Lemma ships agent skills (SKILL.md format) that teach a coding agent how to design, build, and operate pods. lemma skills installs them into the coding agent you already use:

lemma skills list                       # what's bundled
lemma skills install                    # auto-detect Claude Code / Codex / OpenCode / Cursor and install
lemma skills install --target claude    # or pick one explicitly
lemma skills install --all-skills       # include workspace-runtime helpers too

install is an upsert — the CLI owns these skills, so an existing copy is overwritten to match what this lemma-terminal bundles (re-run it after upgrading to refresh them; identical copies report unchanged). By default it installs every namespaced lemma-* product skill: pod building and operation, widgets, app design and QA, research, data analysis, artifact authoring, evaluations, and pod-skill creation. --all-skills additionally installs the environment-specific browser and liteparse-documents helpers. Targets and their locations:

Use --scope project to install into the current directory (.claude/skills/, .agents/skills/, .opencode/skills/, .cursor/skills/), or --dir PATH for an arbitrary location. Cursor is project-scoped only — run lemma skills install --target cursor --scope project inside the repo you're working in. Then restart your coding agent and ask it to build a pod.

How it fits together

• lemma-sdk — the Python client used by functions and automation.
• lemma-terminal (this package) — the human- and agent-facing CLI and TUI; pod-scoped workflows are first-class.
• lemma-stack — installs and manages the local Lemma stack (separate tool).

License

Apache-2.0