Review another agent's change as a Wayglass reviewer agent: list the agent_review queue with wayglass review --for agent --json, read the change with wayglass diff, check it against its declared goal, reason and paths, run its tests, and approve or reject with a reason that says what was checked, bound to the decision reviewed and respecting the no-self-approval and different-model-family rules. Use when the user asks for a reviewer agent, to work through pending_agent changes, to review, check or sign off on another agent's change or a Wayglass intent, or to approve or reject a change in Wayglass. Not for making a change through Wayglass (use the wayglass skill) and never for changes policy routed to a person (pending_human or quarantined).

Make code changes through Wayglass as a coding agent: declare an intent before editing, commit with the why and the Wayglass-Intent trailer, push and act on overlap re-plans, propose and obey the exit code (0 may land, 3 blocked, 4 a person must act), land only an approved change, hand work on with session share and session fork, and check setup with wayglass doctor and per-agent keys. Use when a repository has .wayglass/config.yaml or a wayglass:begin block in AGENTS.md or CLAUDE.md, when the user asks for a change to go through Wayglass, to declare an intent, to propose or land a change, or to run wayglass commands. Not for reviewing another agent's change (use the wayglass-reviewer skill), for writing policy rules, or for deploying the Wayglass Worker.

Build on Cloudflare Artifacts (beta), Git-compatible repositories for agents, through the Workers binding, the REST API and plain git over HTTPS. Use when creating, forking or importing Artifacts repos, minting or revoking repo tokens, cloning or pushing to *.artifacts.cloudflare.net remotes, choosing namespaces and jurisdictions, consuming Artifacts events from Queues or Workflows, or planning around Artifacts limits and pricing. Not for R2 object storage or for Workers Builds connected to an external Git host.

Use the Bubbles v2 component library (charm.land/bubbles/v2, charmbracelet bubbles) inside a Bubble Tea v2 TUI: list, table, viewport, textinput, textarea, spinner, progress bar, filepicker, tree, paginator, timer, stopwatch, help, and key bindings. Use when adding a ready made widget or TUI component (text input, spinners, scrolling view, file picker) to a terminal app, wiring its focus, size, styles, and messages, or migrating v1 bubbles code. Not for the program loop itself (charmed-bubbletea), raw styling (charmed-lipgloss), or whole forms (charmed-huh).

Build terminal user interfaces (TUI) in Go with Bubble Tea v2 (charm.land/bubbletea/v2, the charmbracelet bubbletea framework) using the Elm architecture Model, Update, View loop, commands for I/O, and declarative tea.View fields for alt screen, mouse, cursor, and title. Use when writing or migrating a Bubble Tea or bubbletea app, handling tea.KeyPressMsg or mouse messages, wiring async work into a TUI, or testing one. Not for ready made widgets (charmed-bubbles), styling (charmed-lipgloss), or forms (charmed-huh).

Map of the Charm (charmbracelet) terminal stack and how the pieces fit: which library to reach for (Bubble Tea, Bubbles, Lip Gloss, Huh, Glamour, Log, Wish, Harmonica) or which CLI (gum, vhs, glow, freeze, mods, crush), the v2 charm.land import paths and pinned versions, the dependency order, how the libraries hand data to each other, and a consolidated v1 to v2 migration checklist. Use when starting a TUI or terminal project in Go, choosing a Go TUI stack, deciding between Charm libraries, or planning a charmbracelet v1 to v2 upgrade. For any single library's API use its own charmed-* skill.

Write, review, and refactor Java lambdas, method references, and functional-interface callbacks (Function, Predicate, Supplier, Consumer, Comparator, BiFunction) so they stay readable and behavior-preserving. Use whenever Java code creates or changes a callback passed to a stream, Optional, Map, CompletableFuture, or listener API; when a cleanup or review touches identity lambdas (x returning x), block or multi-line lambdas, method references, suppliers, or side-effecting callbacks; or when the user asks to make Java code cleaner or more functional, even if they never say "lambda". Do not use for stream or Optional semantics themselves, and do not use it to force functional style where a plain loop or branch is clearer.

Bootstraps and uses the canonical LoadComponent host service for Module Federation (src/core/api/services/module-federation/moduleFederationService.tsx). Use whenever the user mentions Module Federation, remoteEntry, LoadComponent, createRemoteComponent, remotes API, bridge components, host shell, dynamic or static remotes, or scaffolding a new host — even if the service file does not exist yet.

Review Java stream performance advice, especially slow stream mappings, external collection mutation with forEach/add, and whether parallelStream is safe; clean up mutation and write or refactor Java Stream and Collector code. Avoid common stream antipatterns such as materializing just to inspect, sorting before min/max, counting for existence, nested stream collections, unsafe null sorting, multi-line lambdas, and careless findFirst/findAny changes. Use whenever writing, reviewing, or refactoring Java code that uses Java streams or collectors, such as stream pipelines, grouping collectors, joining strings, stream first/any element lookup, sorted/limit/distinct stages, primitive stream totals, Optional values inside streams, or parallel streams, including review prompts asking whether a lookup should use findFirst or findAny.

Write, review, and refactor Java Optional code using best practices, improving readability, and preventing common Optional antipatterns such as null-style control flow and readability regressions. Use whenever writing, reviewing, or refactoring Java code that introduces, changes, or reasons about Optional; handles absent, missing, nullable, fallback, or default values where Optional may be appropriate; or touches isPresent/isEmpty, get/orElseThrow, orElse(null), optional.stream(), findFirst/findAny, checked exceptions inside Optional chains, or nullable control flow.

Use when a professional article task needs evidence-bound editorial control: source packet, claim ledger, verification queue, publication notes, anti-slop revision, article audit, or series plan. Trigger on write, rewrite, revise, humanize, outline, audit, or plan for an article, essay, blog post, longform, thought leadership, or source-backed copy. Do not use for generic copy, short posts, email, docs, or fiction. Topic-only => Editorial Brief plus source plan only, no article body. Source packet => article plus Publication Notes. Draft rewrite/revision => first output # Revised Article, then rewritten article body, then Change Notes.

Use when the user wants to design, size, audit, or choose a self-hosted speech recognition or streaming ASR stack, including Whisper, Parakeet, Canary, Riva, NIM, Triton ASR, faster-whisper, sherpa-onnx, voice-agent transcription, Romanian or Moldovan ASR, contact-center transcription, GPU sizing, latency budgets, multilingual routing, VAD, diarization, or production evaluation.

Use when the user wants to review, audit, or check safety for an AI memory system, agent learning pipeline, prompt-tuning workflow, skill builder, trace-mining tool, or eval/feedback loop. Produces an evidence-led audit report with learning-loop map, evidence inventory, maturity scorecard, severity-ranked findings, privacy/provenance gaps, counterfactual/eval coverage, and Stabilize/Standardize/Scale roadmap.

Use when the user wants a landing page audit, CRO review, marketing conversion review, copy audit, funnel alignment check, brand positioning critique, technical performance marketing audit, or prioritized issue report for a landing page, pricing page, signup page, lead magnet, or product page. Runs evidence-first context gathering, parallel specialist analysis, synthesis, and severity-ranked recommendations.

Use when the user wants a test suite audit, test quality or reliability review, regression-protection review, unit/integration/e2e test review, coverage or CI signal assessment, flaky CI investigation, fixture-realism review, spec-drift review, or generated-test validation for AI/LLM/agent-written code. Produces severity-ranked findings for weak assertions, oracle gaps, brittle fixtures, over-mocking, CI trust, and generated-code test risks.

Write or audit AI agent system prompts component-by-component across identity, instruction architecture, behavioral constraints, tools, examples, context strategy, output format, and error handling. Use when the user wants to design a new agent prompt, write a system prompt, review an existing agent prompt, fix tool-use instructions, audit prompt structure, improve context strategy, tune output formats, or define error handling for single-agent or multi-agent systems.