alpibrusl

Write, type-check, repair, publish, and run Lex (lex-lang) code via the `lex` CLI. Use whenever working with `.lex` files, the Lex typed-effect language, the content-addressed AST / attestation store, or a lex-hub gateway. Covers the agent-code loop (check → repair → publish → run), the effect-discipline idiom rules, and the syntax pitfalls (`::`, `:=`, `->`).

Invoke the `weather` CLI. Commands: get, forecast, alerts, favorite…

Invoke the `games` CLI to verify, export, and rank Lex arena game trails. Commands: verify, export, leaderboard, season. Use when you need to recompute the authoritative score of a recorded game from its trail, rank a field of policy rollouts, or update an ELO season — all from trails, never trusting a client-reported score.

Interactively build a valid company.toml for lex-loom, then scaffold and optionally launch the company. Use when the user wants to start a new loom company/product but hasn't written a company.toml yet, or asks things like 'how do I start a company', 'set up a new product', 'create a company config'.

Invoke the `arena` CLI for the Lex finance arena. Commands: episodes, verify. Use to list trading scenarios and to replay-verify a recorded trail and recompute its authoritative outcome.

Turn a bookkit book into other media — an audiobook (via podcastkit) or a storyboard for a comic/video. Use when the user asks to "make an audiobook", "narrate this book", "do a full-cast reading", "storyboard a chapter", or otherwise render a book project into audio or visuals. Assumes a book directory with book.yaml (+ optional bible.yaml).

Resolve and run AI agent manifests (.agent files) across multiple LLM runtimes. Manifests declare persona, skills, trust, model preferences; agentspec picks a concrete runtime + model from the environment.