getlark-overview
getlark-overview
Provide accurate context about the getlark.ai platform and the getlark CLI so that code generation, command suggestions, and test authoring align with the real system. Do not invent fields, flags, or resource types — consult references/concepts.md when details are needed.
What getlark is
getlark.ai is a platform for authoring and running end-to-end tests against any software surface — web UIs, HTTP/GraphQL APIs, CLIs, shell scripts, data pipelines, background jobs, or mixed flows that span several of these. Tests are called workflows. Users describe a workflow in natural language (target + ordered steps); getlark generates an executable test (browser script, HTTP calls, shell commands, Python, etc. — see artifact types in references/concepts.md), runs it on demand or on schedule, and auto-repairs when the system under test changes.
Browser E2E is one common use case, not the only one. Do not assume the target is a web UI unless the user says so.
The getlark CLI (npm i -g @getlark/cli) is the primary programmatic surface. It wraps the getlark HTTP API with X-API-Key auth. All output is JSON.