boardwalk-use-clilisted

# Use the Boardwalk CLI Use this skill whenever the user needs to install, configure, or drive the first-party `boardwalk` CLI, to scaffold a workflow, validate it, sign in, deploy it, trigger a run, cancel one, inspect runs and usage, or manage workflows, secrets, environments, variables, and inference providers. This is the canonical reference for the CLI surface. New to Boardwalk? Read the **`boardwalk-overview`** skill first. It covers what the platform is and the workflow mental model (a workflow is a typed function, not YAML), which this reference assumes you already have. ## What a Boardwalk workflow is A workflow is a **package**: a directory holding `workflow.jsonc` (the deployment descriptor — triggers, permissions, budget, read as data) and an entry exporting a `run` function the platform calls — `src/index.ts` with `export default async function run(input, context)`, or `main.py` with a module-level `async def run`. There is no YAML pipeline and no DSL. Native types on the signature are the I/O contract (the deploy derives their schemas for the dashboard's run form), and the return value is the run's output. The body calls `agent(prompt)` for LLM work (`model` optional, chosen per call) and durable primitives (`secrets.get`, `sleep`, `workflows.call`, `humanInput`) for everything else. See `boardwalk-overview` for the full model, `write-good-workflows` for authoring it well, and `equip-agents` for giving an `agent()` skills, tools, MCP, and memory. **Write pl