the catalog

102 specifications, 1828 graded capabilities

Every spec is a versioned tree of capabilities. Read one, build against it, then submit a grade — free, no account.

See two implementations graded against one spec →

specs.json — the whole catalog as one file, for agents.

apps 18

Calculator Arithmetic that behaves the way a pocket calculator does, including the awkward parts. 17 capabilities · reference Calendar Events, recurrence and invitations — where time zones go to fight. 20 capabilities · reference Chat App Rooms, message history, presence and delivery state for a realtime chat client. 16 capabilities · reference Email Client Read, write and organise mail — offline, threaded and quick. 20 capabilities · reference Excalidraw Hand-drawn-style whiteboard: shapes, freedraw, selection, export and live collaboration. 15 capabilities · extracted Expense Tracker Categorised spending with budgets, recurring entries and monthly totals. 14 capabilities · reference Kanban Board Columns and draggable cards with WIP limits, filtering and persistence. 17 capabilities · reference Markdown Editor Live preview, formatting shortcuts, and a clean round-trip to disk. 14 capabilities · reference Music Player Library, queue and playback that behaves when the network or the file does not. 18 capabilities · reference Note-Taking App Capture quickly, find later, and never lose a note to a sync conflict. 18 capabilities · reference Obsidian (desktop) Local-first markdown vault: wikilinks, backlinks, graph view and a plugin surface. 19 capabilities · extracted Parity (the ProveSpec workbench) The application that measures whether one implementation has what another declares — specified against itself. 34 capabilities · extracted Password Manager Hold every credential — where the security model is the product. 18 capabilities · reference Photo Library Thousands of images: import, organise, find, and never lose the original. 19 capabilities · reference Pomodoro Timer Work/break cycles with notifications, skip/reset and session history. 12 capabilities · reference RSS Reader Feed subscriptions, unread tracking, folders and offline reading. 14 capabilities · reference TodoMVC The canonical todo app — the reference specification every framework implements. 20 capabilities · reference Weather Dashboard Current conditions, multi-day forecast, saved locations and units. 13 capabilities · reference

engines 12

Cron Scheduler Run things on time — including across daylight saving and restarts. 17 capabilities · reference Diff Engine Compute and apply differences between two texts. 15 capabilities · reference Image Processing Engine Decode, transform and encode images without wrecking them. 17 capabilities · reference JSON Parser Parse and serialise JSON per RFC 8259, with useful errors. 15 capabilities · reference KV Store Get/put/delete, iteration, atomic batches, durability and compaction. 12 capabilities · reference Markdown Parser CommonMark-shaped parser: text in, structured document or HTML out. 18 capabilities · reference PDF Renderer Draw a PDF faithfully, quickly, and without trusting its contents. 18 capabilities · reference Regular Expression Engine Pattern matching with stated semantics and no catastrophic surprises. 17 capabilities · reference SQLite Embedded SQL engine: transactions, indexes, WAL, triggers and extensions. 18 capabilities · extracted State Machine Engine States, events and transitions — with the illegal moves impossible. 17 capabilities · reference Task Scheduler (in-process) Run work concurrently with bounded resources and predictable ordering. 16 capabilities · reference Template Engine Render templates against data: interpolation, control flow, partials, escaping. 16 capabilities · reference

services 10

API Gateway One front door to many services: routing, auth, limits and observability. 17 capabilities · reference Auth Service Registration, sessions, password reset, OAuth providers and role checks. 15 capabilities · reference Feature Flag Service Evaluate flags per user with targeting, rollouts and audit. 18 capabilities · reference File Upload Service Multipart uploads, validation, storage backends and signed URLs. 13 capabilities · reference Job Queue Enqueue background work and run it reliably: retries, scheduling, dead letters. 19 capabilities · reference Notification Service Deliver the right message on the right channel without becoming noise. 18 capabilities · reference PocketBase Single-binary backend: SQLite records, realtime subscriptions, auth, files and hooks. 14 capabilities · extracted URL Shortener Slug generation, redirects, click analytics, expiry and custom domains. 19 capabilities · reference Web Analytics Service Measure usage without becoming the thing users install a blocker for. 16 capabilities · reference Webhook Relay Receive, verify, queue and deliver webhooks with retries and replay. 18 capabilities · reference

devtools 8

Bundler Turn a module graph into deployable assets, quickly and reproducibly. 18 capabilities · reference Code Formatter One canonical layout, applied without argument. 16 capabilities · reference Container Runtime Build, run and isolate application images reproducibly. 16 capabilities · reference Debugger Stop a running program at the right moment and see why it is wrong. 18 capabilities · reference Linter Find problems in source without running it, without crying wolf. 18 capabilities · reference Package Manager Resolve, install and lock dependencies so two machines agree. 17 capabilities · reference Test Runner Run tests, report failures usefully, and stay fast enough to run often. 18 capabilities · reference Version Control System Track history, branch, merge — and never lose committed work. 18 capabilities · reference

formats 8

CSV (RFC 4180) Comma-separated values — the format everyone thinks is simple. 16 capabilities · reference iCalendar The calendar interchange format — recurrence, time zones and the parts that break sync. 16 capabilities · reference JSON Web Token Signed claims in a URL-safe string — and the ways it goes wrong. 15 capabilities · reference Markdown with Front Matter The content format behind most static sites — document plus structured metadata. 14 capabilities · reference OpenAPI Description Describe an HTTP API precisely enough to generate clients and tests from it. 16 capabilities · reference Semantic Versioning 2.0.0 Version numbers that carry a promise about compatibility. 15 capabilities · reference TOML A configuration format designed to be unambiguous — including the parts that catch people. 16 capabilities · reference YAML 1.2 Human-friendly data serialisation — indentation, anchors and famous footguns. 15 capabilities · reference

infra 7

Cache Server In-memory key/value cache: eviction, expiry and what happens when it fills. 17 capabilities · reference Content Delivery Network Serve bytes close to the user, and invalidate them when they change. 17 capabilities · reference Load Balancer Spread traffic across healthy backends without dropping requests. 16 capabilities · reference Message Broker Publish, route and consume messages with the delivery guarantee written down. 17 capabilities · reference Object Storage Buckets and objects with the consistency and lifecycle rules stated. 17 capabilities · reference Search Index Full-text search: analysis, ranking, and answers that arrive fast. 17 capabilities · reference Secrets Manager Store, distribute and rotate credentials without them landing in a file. 16 capabilities · reference

protocols 6

GraphQL API A typed query interface where the client states exactly what it needs. 16 capabilities · reference HTTP Caching Freshness, validation and invalidation between client, proxy and origin. 16 capabilities · reference OAuth 2.1 Authorization Code + PKCE Delegated authorisation done the way current guidance requires. 17 capabilities · reference SMTP Submission Handing an email to a server so that it arrives rather than lands in spam. 16 capabilities · reference WebRTC Peer-to-peer audio, video and data — including getting through the NAT. 16 capabilities · reference WebSocket A persistent bidirectional connection, from handshake to graceful close. 16 capabilities · reference

games 5

2048 Sliding tile puzzle: merge equal tiles on a 4×4 grid until 2048 appears. 17 capabilities · reference Conway's Game of Life Cellular automaton: cells live, die and reproduce by neighbour count. 16 capabilities · reference Minesweeper Uncover every safe cell on a mined grid using neighbour counts. 17 capabilities · reference Snake Grow a snake by eating food without hitting the wall or yourself. 16 capabilities · reference Tetris Falling tetrominoes — where the modern rules are far more specific than people expect. 19 capabilities · reference

cli 4

HTTP Client CLI Make HTTP requests from the terminal and read the answer without effort. 17 capabilities · reference JSON Tool CLI Query, filter and reshape JSON from a pipe, with exit codes that mean something. 12 capabilities · reference ripgrep Recursive regex search with gitignore awareness, type filters and fast output. 15 capabilities · extracted Todo CLI Manage a task list from the terminal, with output humans and scripts can both read. 16 capabilities · reference

commerce 4

Checkout Turn a cart into a paid order without losing the customer or double-charging them. 19 capabilities · reference Invoicing Documents that have to be legally correct, sequential and impossible to quietly alter. 16 capabilities · reference Shopping Cart Hold items, keep the total honest, and survive the customer leaving and coming back. 18 capabilities · reference Subscription Billing Recurring charges, plan changes and the proration nobody gets right first time. 19 capabilities · reference

editors 4

Code Editor Text editing for code: navigation, completion, diagnostics, multi-file work. 21 capabilities · reference Diagram Tool Boxes, arrows and layout that survives being edited. 18 capabilities · reference Presentation Editor Slides, layouts and a presenter mode that works on someone else’s projector. 19 capabilities · reference Spreadsheet Cells, formulas and recalculation — the part everyone underestimates. 20 capabilities · reference

goods 4

Coffee Grinder A deliberately non-software spec: same tree, same grading, no code anywhere. 19 capabilities · reference Mattress A third non-software spec — the domain has no engine and no interface. 18 capabilities · reference Office Chair What actually differs between chairs, once marketing is removed. 18 capabilities · reference Running Shoe The second non-software example — proof the tree is domain-agnostic. 18 capabilities · reference

hardware 3

Laptop What to compare when comparing portable computers. 21 capabilities · reference Smartphone The comparison people actually make before spending money. 20 capabilities · reference Solid-State Drive Beyond the headline sequential number. 17 capabilities · reference

libraries 3

Date & Time Library Instants, zoned times, durations and formatting — correct across zones and DST. 18 capabilities · reference Shell (TS workbench) Reusable VS Code-style shell: activity bar, sidebars, toolbar, status bar, toasts. 22 capabilities · extracted Validation Library Declare a schema, validate unknown input, get typed data or precise errors. 20 capabilities · reference

ports 3

nanochat (Python) Karpathy's nanochat — the reference implementation: tokenizer, pretraining, fine-tuning, RL and inference. 36 capabilities · extracted NanoChat Web Runtime The browser runtime of nanochat, graded against the same Python reference. 36 capabilities · extracted NanoChat.NET The .NET port of nanochat, graded capability by capability against the Python reference. 38 capabilities · extracted

layers 2

NanoChat.NET — engine layer The .NET port's engine capabilities, as reachable from code. 12 capabilities · extracted NanoChat.NET — UI surface What the desktop UI actually exposes of that same engine. 49 capabilities · extracted

generators 1

Hugo Static site generator: content model, templates, taxonomies and asset pipelines. 14 capabilities · extracted