Documentation
¶
Overview ¶
Command flipper-usecases runs realistic operator tasks against a live Flipper Zero and reports pass/fail + concise summaries.
Complementary to flipper-validate: that binary exercises every primitive one-by-one to prove the transport works. This binary exercises the *user-visible* surface — the short natural-language requests operators actually type into PromptZero ("scan my fob", "what's on my Flipper", "listen for garage doors on 433") — and shows what the tool_result would look like when the agent dispatches them. No LLM required; each use case is a thin operator-prompt → tool-call mapping.
Use this to catch "the tool returns success=true but didn't do what the operator wanted" regressions before they reach a real engagement. See docs/RELEASING.md for how results feed the Verified block in CHANGELOG.md.
Source Files