⚠️ This repo is in development right now. ⚠️

Lag CLI: Lamin Agent CLI

lag can execute existing runnable tools and author/update runnable scripts (.py).

Setup

pip install lag-cli

Create ~/llms.env with:

GEMINI_API_KEY=your_api_key_here

Make sure LaminDB is initialized and connected.

Quick Start

Run in default mode:

lag --prompt "Write a text file with 'Hello agent!' in it, please"

You can explore runnable example scenarios in tests/examples.

Modes

Default mode (execute only)

• If a tool exists (tool.md or latest tool_*.md, or --tool-file), lag executes the referenced runnable tools.
• Otherwise, lag executes existing runnable tools referenced in --prompt (explicit .py key or path).
• Default mode does not create or update tools. If a referenced tool is missing, it fails with a clear error.

Planning mode (--tool)

Generate or update runnable tools (without executing them):

lag --tool --prompt "Update test-lag/create_fasta.py with another protein"

Generated/updated tool files are saved via lamin save in tool mode.

Common Flags

• --project <name> sets LAMIN_CURRENT_PROJECT.
• --model <model-name> selects the Gemini model (--tool mode).
• --output-file <path> sets output filename for generated content (--tool mode).
• --tool-file <path> executes a specific tool in default mode.
• --no-track disables ln.track() / ln.finish() injection in generated scripts (--tool mode).

Run Context Propagation

When lag executes scripts, it propagates:

• LAMIN_INITIATED_BY_RUN_UID (master run uid)
• LAMIN_CURRENT_PROJECT (if --project is provided)

lag itself does not directly create output artifacts; produced outputs are tracked by executed tool code.