@acoyfellow/lab-cli

Minimal CLI for Lab. Agents shell out, get traces back.

Install

npm install -g @acoyfellow/lab-cli

Usage

# Run JS in a single isolate
lab run 'return 1 + 1'

# Run a multi-step chain
lab chain '[{"body":"return [1,2,3]","capabilities":[]},{"body":"return input.map(n=>n*2)","capabilities":[]}]'

# Spawn nested isolates
lab spawn 'const a = await spawn("return 10", []); return a'

# AI generates code from a prompt, then runs it
lab generate 'Sum the numbers 1 to 10'

# Fetch a stored trace
lab trace abc123

# Seed demo KV data
lab seed

Every command prints JSON to stdout. Every run includes a traceId pointing to the permanent trace.

Environment

| Variable | Default | Description | |----------|---------|-------------| | LAB_URL | http://localhost:1337 | Lab Worker origin |

For agents

An agent shells out to lab, reads the JSON result, and uses the traceId as proof of execution:

result=$(lab chain '[{"body":"return 42","capabilities":[]}]')
traceId=$(echo "$result" | jq -r '.traceId')
# traceId is now a permanent, inspectable artifact

Or from a script:

lab run 'return { computed: Math.pow(2, 32) }' | jq '.result'

Built with Effect

Internals use Effect v4 for structured error handling. The CLI interface is just stdin/stdout — Effect stays inside.

License

MIT