saschb2b

Create, inspect, author, revise, enrich, audit, repair, research, migrate, validate, and maintain knowledge for AI agents as conformant Open Knowledge Format (OKF) bundles. Use for /okf commands; work on an existing OKF bundle; documentation of tables, datasets, views, metrics, runbooks, playbooks, APIs, schemas, or join paths; exports from catalogs, metadata, websites, URLs, or source code; and any knowledge an agent will read later. Apply implicitly when editing an OKF bundle so conformance, timestamps, log.md, indexes, citations, and cross-links remain current.

Adopt the structural operating discipline of Anthropic's Fable model in whatever agent harness is running, so the host model (GPT 5.6, Gemini, or any coding agent) orients, parallelizes, delegates, verifies, and reports the way Fable does. Ships the doctrine as a vendored OKF bundle; the operating loop, parallel dispatch, delegation economy, orchestration patterns, verification gates, faithful reporting, the irreversibility gate, context economy, tool and code conduct, memory discipline, trust boundaries against prompt injection, a harness-mapping runbook, and a delta map against the leaked GPT 5.6 system prompt. Use when the user asks the agent to behave or think like Fable or Claude, act more structurally like Fable, enter fable mode, work like an Anthropic agent, tighten agentic workflow discipline, batch tool calls, verify before claiming done, stop narrating and start acting, or invokes /fable-mode. This changes how work is executed; choosing the work autonomously stays with autopilot.

Detect test-suite quality smells in any language or framework, explain the cost of each, and propose a fix with a source link. A counterweight to agents that blindly chase coverage and ship huge, slow, flaky suites. Covers test intent and naming, assertions (assertion-free, weak, tautological, snapshot-everything), fixtures and duplication (DAMP over DRY, mystery guests, logic in tests), test doubles (over-mocking, mock drift), determinism (time, randomness, order, shared state, hermeticity), async waiting (sleeps vs condition waits), speed and cost (inverted pyramid, redundant layers), coverage theater, and lifecycle hygiene (skipped-test rot, retries as policy, green-by-rerun), plus a suite-level shape pass with a slimming plan and a mechanical gate stack. Use when reviewing or writing tests, auditing, slimming, or speeding up a test suite, hunting flaky or slow tests, or asked to sniff, smell-check, or clean up specs or a test folder. Defers production-code smells to react-stinky and tauri-stinky.

Author, validate, and maintain knowledge for AI agents as conformant Open Knowledge Format (OKF) bundles, Google's vendor-neutral spec (v0.1) of markdown files with YAML frontmatter for portable agent context. Provides commands init, add, enrich, link, index, log, validate, export, and consume over a bundle. Use when the user types /okf or asks to start an OKF bundle, document a BigQuery table, dataset, view, metric, runbook, playbook, API, or join path so an agent can read it, make internal knowledge agent-readable or portable, export a data catalog, metadata, or webpages and external URLs to OKF, or follow the OKF spec. Also apply implicitly whenever you write, transform, or export knowledge an agent will later read (table and metric docs, runbooks, deprecation notices, schemas, docstrings), so it ships as a conformant bundle with a type field per concept instead of ad-hoc prose, and whenever you edit an existing OKF bundle, keeping it conformant by refreshing timestamps, log.md, index.md, and cross-links.

Detect Rust and Tauri maintainability smells across the backend crate, the IPC boundary, and the app shell, explain the cost of each, and propose a fix with a source link. Covers Rust discipline (error handling and panics, ownership and clones, type design, module structure, async correctness, overgeneration and dead code, unsafe and lint hygiene, testing) and the Tauri layer (command design, managed state, events vs channels, capabilities and security, IPC contract and type codegen, config and build footprint), plus a cross-boundary contract audit and a mechanical gate stack (fmt, clippy, test, audit, deny, machete). Use when reviewing or writing Rust code or a Tauri app, auditing src-tauri, setting up quality gates or CI lints for Rust, or asked to sniff, smell-check, lint, or clean up a Rust or Tauri codebase, crate, command, or pasted snippet. Defers React component and hook smells to react-stinky and color literals to theme-colors.

Detect React and TypeScript maintainability smells across the whole component, hook, and module, explain the cost of each, and propose a fix with a source link. Covers prop and API design (naming, booleans, callbacks, variants, controlled state, generics, refs, styling, JSDoc), state and data flow (redundant state, stale closures, context and ref misuse), effects and lifecycle, component structure and hooks, rendering correctness (keys, mutation, impure renders, hydration, the leaked 0 from &&), accessibility and focus management, async handlers, error boundaries, unsafe HTML, TypeScript discipline (any, casts, untyped catch, exhaustiveness), and a cross-file duplication pass. Defers memoization and re-render performance to react-compiler and color literals to theme-colors. Use when reviewing or writing React components or hooks, auditing for maintainability, or asked to sniff, smell-check, lint, or clean up a codebase, folder, file, or pasted snippet. Respects native HTML attributes and library conventions.

Default to the latest stable major versions of JavaScript and TypeScript frameworks and their current paradigms, instead of the older versions and patterns that dominate LLM training data. A version-and-paradigm changelog index that routes to per-tool notes. Use when writing, reviewing, scaffolding, or upgrading any JS or TS project, when choosing dependency versions, when an agent emits an outdated pattern, or when working with React, Angular, Vue, Svelte, Solid, Next.js, Nuxt, React Router, Remix, SvelteKit, Astro, TanStack Start, Material UI, Tailwind, shadcn, Mantine, Chakra, TanStack Query, Redux, Zustand, Jotai, TypeScript, Vite, Node, ESLint, pnpm, Bun, or Biome. Check the project's installed version first; this snapshot ages.

Write and revise human-facing prose in a plain, precise, professional register, stripping the tells that mark text as AI-generated. Use when writing or editing a README, blog post, doc, PR or commit message, email, release note, changelog, or any copy a person will read. Also use when the user asks to make text sound less like AI, remove AI slop, cut the marketing voice, reduce over-structuring or list-itis, tighten, de-fluff, or professionalize a draft, or when reviewing prose you just generated before handing it over. Extends to writing inside code. Code comments, commit messages, PR descriptions, test names, error and log messages, and markdown formatting like headings, bold, emoji, and badges.

Work autonomously on a project without waiting for direction. On command, survey the codebase, pick one high-value low-risk improvement, do it, verify it, commit it, then pick the next, looping like a self-directed intern. Use when the user hands off open-endedly ("work on your own", "keep improving this", "find something worth doing", "don't wait for me", "go autonomous", "run for a while", "be a passionate intern", "do whatever you think is valuable") or invokes /autopilot, and you have to choose the work yourself rather than wait for instructions or feedback.

Author, validate, and maintain knowledge for AI agents as conformant Open Knowledge Format (OKF) bundles, Google's vendor-neutral spec (v0.1) of markdown files with YAML frontmatter for portable agent context. Provides commands init, add, enrich, link, index, log, validate, export, and consume over a bundle. Use when the user types /okf or asks to start an OKF bundle, document a BigQuery table, dataset, view, metric, runbook, playbook, API, or join path so an agent can read it, make internal knowledge agent-readable or portable, export a data catalog, metadata, or webpages and external URLs to OKF, or follow the OKF spec. Also apply implicitly whenever you write, transform, or export knowledge an agent will later read (table and metric docs, runbooks, deprecation notices, schemas, docstrings), so it ships as a conformant bundle with a type field per concept instead of ad-hoc prose, and whenever you edit an existing OKF bundle, keeping it conformant by refreshing timestamps, log.md, index.md, and cross-links.

Write and review React code as if the React Compiler is enabled. Skip manual useMemo, useCallback, and React.memo by default, and audit existing code for stale manual memoization. Use when writing or refactoring any React component in a React 19+ codebase, when reviewing a React PR, when setting up the eslint-plugin-react-hooks configuration, or when seeing memoization patterns in code an LLM produced. Counterweight to pre-compiler React in LLM training corpora, which biases default output toward manual memoization unless prompted otherwise.

Detect React and TypeScript maintainability smells across the whole component, hook, and module, explain the cost of each, and propose a fix with a source link. Covers prop and API design (naming, booleans, callbacks, variants, controlled state, generics, refs, styling, JSDoc), state and data flow (redundant state, stale closures, context and ref misuse), effects and lifecycle, component structure and hooks, rendering correctness (keys, mutation, impure renders, hydration, the leaked 0 from &&), accessibility and focus management, async handlers, error boundaries, unsafe HTML, TypeScript discipline (any, casts, untyped catch, exhaustiveness), and a cross-file duplication pass. Defers memoization and re-render performance to react-compiler and color literals to theme-colors. Use when reviewing or writing React components or hooks, auditing for maintainability, or asked to sniff, smell-check, lint, or clean up a codebase, folder, file, or pasted snippet. Respects native HTML attributes and library conventions.

Audit and refactor UI code so every color value comes from the theme. Replace hex codes, rgba(), hsl(), oklch() literals, and named CSS colors with theme palette tokens (primary, secondary, error, warning, info, success, background, text) and the alpha() helper for transparency. Use when writing or reviewing any UI component, when starting a new MUI or design-system theme, when a designer hands off colors as hex annotations, when seeing color literals scattered across component files, or when an LLM produced styling with hardcoded hex values. Counterweight to LLM training corpora dominated by tutorial code that inlines hex codes and rgba literals instead of role-based theme tokens.

Work autonomously on a project without waiting for direction. On command, survey the codebase, pick one high-value low-risk improvement, do it, verify it, commit it, then pick the next, looping like a self-directed intern. Use when the user hands off open-endedly ("work on your own", "keep improving this", "find something worth doing", "don't wait for me", "go autonomous", "run for a while", "be a passionate intern", "do whatever you think is valuable") or invokes /autopilot, and you have to choose the work yourself rather than wait for instructions or feedback.

Force UX questioning before any UI work, refactor, or redesign. Use when the user asks to design, redesign, restyle, polish, "fix", or "make nicer" a screen, component, page, or flow. Also use when the user describes a problem in UI terms ("the button is hard to find", "this form is confusing", "this page is ugly") and is about to reach for a visual fix, since most UI-shaped complaints have UX-shaped causes.

When the user wants to audit, review, or diagnose SEO issues on their site. Also use when the user mentions "SEO audit," "technical SEO," "why am I not ranking," "SEO issues," "on-page SEO," "meta tags review," or "SEO health check." For building pages at scale to target keywords, see programmatic-seo. For adding structured data, see schema-markup.

Detect and fix visual and layout consistency defects in rendered UI, the systematic raggedness AI generators emit and seniors notice at a glance. Covers uneven card heights in a grid, sibling cards whose internal sections (title, body, footer) do not line up, trailing icons such as open-in-new-tab at ragged X positions, and arbitrary off-scale spacing, plus alignment, repeated elements (buttons, radius, shadow, icons), type scale, tables and numeric alignment, borders, overflow, layout shift, touch targets, and focus rings. Each smell carries an objective detection signal, a concrete CSS fix (subgrid, equal-height grids, spacing tokens, tabular-nums), an exception guard, a severity, and whether an agent may fix it freely or must surface a design call first. Use when reviewing or building any UI, when a card grid or list looks busy or ragged, when spacing feels arbitrary, or when asked to tidy or align a layout so it scans cleanly. Defers color to theme-colors and component code smells to react-stinky.

Modern Android development with Jetpack Compose as the default UI toolkit and Material 3 Expressive as the default design direction, a dated snapshot. Covers Compose fundamentals (composables, modifiers, state hoisting, side effects, lazy lists, strong skipping, recomposition), Material 3 and Expressive theming and components (MaterialExpressiveTheme, MotionScheme, button groups, split buttons, FAB menu, floating toolbars, wavy progress indicators, shape morphing, dynamic color), architecture (ViewModel, StateFlow, collectAsStateWithLifecycle, Hilt), type-safe Navigation Compose and Navigation 3, data (Room, DataStore, Retrofit, Ktor, Coil), the Gradle Kotlin DSL build with version catalogs and KSP, testing, performance, and View interop. Use when building Android UI in Compose, theming with Material 3 or Expressive, wiring architecture or navigation, configuring the build, or debugging Compose issues. Check the project's versions first; re-verify version facts against the official Android and AndroidX docs.

Write and revise human-facing prose in a plain, precise, professional register, stripping the tells that mark text as AI-generated. Use when writing or editing a README, blog post, doc, PR or commit message, email, release note, changelog, or any copy a person will read. Also use when the user asks to make text sound less like AI, remove AI slop, cut the marketing voice, reduce over-structuring or list-itis, tighten, de-fluff, or professionalize a draft, or when reviewing prose you just generated before handing it over.