lev-os/agents

tla-plus

Use when learning or applying the TLA+ ecosystem end-to-end: selecting VS Code, Toolbox, or CLI; choosing PlusCal vs raw TLA+; writing specs above the code level; structuring properties and cfg files; running SANY, TLC, Apalache, or TLAPS; debugging…

Idioma do texto original: inglês

testing-real-service-e2e-no-mocks

Build mock-free integration and E2E tests that hit real databases, real APIs, and real services with structured logging. Use when: replacing mocked tests with real-DB tests, transaction rollback isolation, test data factories, structured JSON-line test…

Idioma do texto original: inglês

now

Compose evidence-backed pages (lessons, technical explainers, run reviews, sales letters, dashboards, feedback surfaces, document collections) as single-file HTML in the lev.now shell. By default, write a content-specific GenUI page, inspect it in light and…

Idioma do texto original: inglês

browser-skill

Automate the user's logged-in Chromium browser: read pages, fill forms, scrape data, operate tabs, test a UI, or debug a website. Requires the bsk CLI and browser extension.

Idioma do texto original: inglês

interview

Use when guiding decision interviews that produce a design artifact in .lev/pm/designs before proposal or implementation planning.

Idioma do texto original: inglês

lev-plan

Use when compiling captured or designed intent into a Lev plan (human runbook, migration, architecture plan, multi-slice roadmap), deepening or reviewing a plan, changing plan state under `.lev/pm/plans/`, reviewing slice readiness, rendering a slice map, or…

Idioma do texto original: inglês

prior-art

Use when you need an evidence-first provenance report from `cass`, repo docs, `.lev`, `~/.agents/diagrams`, `qmd`, or Grep/Glob tools.

Idioma do texto original: inglês

work

Use when routing tracked work through lifecycle lanes, resolving workstream context, or deciding which lifecycle skill owns the next entity transition.

Idioma do texto original: inglês

ll

Typed tick loops with FlowMind validation and parallel tick dispatch. Use when running /ll, recurring pipelines, quality loops, or autonomous scheduler ticks.

Idioma do texto original: inglês

capture

Use when inventorying conversation state into durable artifacts with an operator-visible source-fidelity table, routing faithful items toward interview, planning, proposal, or execution, or deep-processing a brain dump.

Idioma do texto original: inglês

lev

Use when routing intent into Lev lifecycle work, clarifying an uncertain next step, or invoking and diagnosing the Lev CLI and runtime.

Idioma do texto original: inglês

agent-desktop

Operate desktop applications on macOS, Windows and Linux through native accessibility trees, using the agent-desktop CLI. Use when an AI agent must see or act on a desktop GUI: click buttons, fill forms, read UI state, navigate menus, toggle checkboxes,…

Idioma do texto original: inglês

arch

Architecture reasoning with trade-offs, ADRs, C4, and fitness functions. Use when user asks for system design, tech selection, architecture review, or ADR drafts.

Idioma do texto original: inglês

dump

Use when a long brainstorm or brain dump needs deep capture, prior-art routing, and durable next states.

Idioma do texto original: inglês

horizon

Use when the operator explicitly invokes /horizon or skill://horizon for a long-horizon planning brief, or when lev-plan loads the method as its conditional planning reference.

Idioma do texto original: inglês

idea

Use when shaping an idea with the wizard, synthesizing related concepts, defining a candidate ontology, or resuming an idea in .lev.

Idioma do texto original: inglês

propose

Use when compiling aligned simple intent or a source-faithful Lev plan into a reviewable slice map or one execution-ready Lev task. Compatibility alias for lev-plan; it does not create or repair broad plans.

Idioma do texto original: inglês

research

Use when current external facts, source-backed comparisons, investigations, or recent community evidence are needed.

Idioma do texto original: inglês

pr-triage

Use when triaging a scoped set of pull requests as a merge train - gate each PR for readiness, bounce what is not ready, merge what is ready into one integration branch, then review the whole result once with two models and an architect, fix blocking…

Idioma do texto original: inglês

sync

Checkpoint and synchronize the current Git checkout and dirty submodules using Git directly. Semantically merge clear conflicts and ask only when intent is unresolved. Default scope is all dirty work; paths after -- narrow it. --handoff returns a copyable…

Idioma do texto original: inglês

codebase-memory

Use the codebase knowledge graph for structural code queries. Triggers on: explore the codebase, understand the architecture, what functions exist, show me the structure, find code from plain words, search by meaning, who calls this function, what does X…

Idioma do texto original: inglês

computer-use

Use when working with macOS GUI apps, desktop automation, background clicks, drag or canvas tasks. Agents run JavaScript in one open session through scripts/cu.mjs (code mode, no MCP registration); the engine is OpenAI's Codex Computer Use when the ChatGPT…

Idioma do texto original: inglês

tools

Atlassian, Rovo, Robo, Teamwork Graph, TWG, Jira, Confluence; REA reverse engineering, MCP-to-CLI integration; GitHub, gh, issues, pull requests; Notion, ntn, pages, databases; Slack, chat, messaging, Block Kit; Exa, web search, semantic search, contents;…

Idioma do texto original: inglês

live

Maintain a local JSON-backed progress view with live refresh, nested work lists and linked evidence. Use for browser status reports and long-running work dashboards; execution stays with the existing work owner.

Idioma do texto original: inglês

ux

Use when designing UX/product flows, researching users or personas, conducting Agentic user research with day-in-the-life simulations, creating wireframes or IA, working with Pencil visuals, or designing agentic CLI UX.

Idioma do texto original: inglês

skill-discovery

Use when a task, agent brief, UX route, CDO plan, wave, or sprint needs relevant local skill material, or when selected material's applicability and loading must be checked.

Idioma do texto original: inglês

orch

Use when coordinating a multi-step goal across agents, skills, Lev lifecycle routes, scheduling, research, implementation, review, or human decisions.

Idioma do texto original: inglês

auto-enrich

Use when source-faithful planning artifacts need simple host-only, standard, or deep enrichment before SDLC proposal review or a non-coding domain handoff.

Idioma do texto original: inglês

poc

Use when the user asks to prove, spike, prototype, POC, validate a production path, compare landing zones, or decide whether an idea should move to propose, monitor, reject, or archive.

Idioma do texto original: inglês

close

Use when verified work is ready for QA, acceptance, checkpointing, learning, recommendation, and handoff.

Idioma do texto original: inglês

coder

Run and supervise Lev-aware coding workers across Codex, Claude Code, Gemini, OpenCode, Pi, OmO Native (omo), or oh-my-pi (omp), including CAAM identity handling and single-worker or coder-reviewer SDD modes.

Idioma do texto original: inglês

exec

Use when running execution-ready task slices, dispatching verified `lev exec` flows, or reviewing implementation work in the Exec lane.

Idioma do texto original: inglês

goal-exec

Convert an explicit user request into a Codex goal and pass its chosen coder, Lev, omo/omp, SDD, or Fable execution flags to $coder.

Idioma do texto original: inglês

ultraqa

Use when adversarially reviewing behavior with dynamic e2e evidence: hostile scenario matrices, misleading-success and harness-error checks, bounded QA cycles, cleanup evidence, and repair requests returned to the caller.

Idioma do texto original: inglês

creative

Use when selecting UX research, wireframes, UI implementation or review, mobile design, visual direction, graphics, animation, or video methods.

Idioma do texto original: inglês

accessibility-expert

Build inclusive interfaces with WCAG compliance and best practices.

Idioma do texto original: inglês

ai-image-generation

Generate AI images with GPT-Image-2, FLUX, Gemini, Grok, Seedream, Reve and 50+ models via inference.sh CLI. Models: GPT-Image-2, FLUX Dev LoRA, FLUX.2 Klein LoRA, Gemini 3 Pro Image, Grok Imagine, Seedream 4.5, Reve, ImagineArt. Capabilities: text-to-image,…

Idioma do texto original: inglês

ai-native-product-designer

AI Native Product Designer role - LLM-first workflow, AI code prototyping, Figma as finish line, self-serve research, outcome ownership. Use for leveling, workflow design, and AI readiness reviews. Includes quick reference, full rubric appendix, checklist,…

Idioma do texto original: inglês

ai-product

AI product engineering for shipping LLM features to production.

Idioma do texto original: inglês

ai-product-strategy

Help users define AI product strategy. Use when someone is building an AI product, deciding where to apply AI in their product, planning an AI roadmap, evaluating build vs buy for AI capabilities, or figuring out how to integrate AI into existing products.

Idioma do texto original: inglês