pax-k

Diagnose failed or unreliable agent workflows by classifying bad assumptions, missing observations, tool misuse, wrapper failures, hidden repair loops, incomplete validation, handoff gaps, and repeated failure patterns. Use after a failed harness run, flaky agent output, or confusing validation loop.

Review agent-generated changes against project rules. Use when asked for a review, before opening a PR, after large Codex edits, or when checking diffs for correctness, security, architecture, tests, docs drift, dependencies, migrations, and release risk.

Generate or maintain a minimal high-signal AGENTS.md for coding agents. Use whenever a repo needs agent instructions, an existing AGENTS.md is too long or vague, or project docs/scripts need a concise routing map for Codex.

Define and mechanically enforce architecture boundaries for a repo. Use when a project needs allowed dependency rules, forbidden imports, layer documentation, dependency maps, circular dependency checks, or protection against agent-created architectural drift.

Scaffold a new TypeScript web, mobile, API, or full-stack project with Better-T-Stack, then harden it with the Pax Fullstack Harness. Use whenever the user wants a new TypeScript app scaffold, mentions Better-T-Stack, asks for a modern type-safe stack, or wants Codex to start a repo that is agent-ready from day one.

Use guided Chrome DevTools MCP setup and workflows for browser runtime debugging. Use when a web app needs console, network, DOM, accessibility, Lighthouse, performance, responsive, or live-browser investigation beyond normal tests.

Build PR validation, local CI reproduction, CI triage, and deployment gates. Use when a repo needs GitHub Actions beyond basic validation, coverage/artifact expectations, branch protection notes, CI failure triage docs, or release/deployment readiness gates.

Inventory and validate user-facing browser workflows with click-path evidence, accessible controls, Playwright coverage, console/network checks, and lifecycle verification. Use for UI work, web regressions, shadcn/ui changes, todo-style create/edit/delete flows, or browser acceptance testing.

Ingest documents into the Obsidian wiki by distilling their knowledge into interconnected wiki pages. Use this skill whenever the user wants to add new sources to their wiki, process a document or directory, import articles, papers, or notes into their knowledge base, or says things like "add this to the wiki", "process these docs", "ingest this folder". Also triggers when the user drops a file and wants it incorporated into their existing knowledge base. Also handles raw mode: "process my drafts", "promote my raw pages", or any reference to the _raw/ staging directory.

Answer questions by searching the compiled Obsidian wiki. Use this skill when the user asks a question about their knowledge base, wants to find information across their wiki, asks "what do I know about X", "find everything related to Y", or wants synthesized answers with citations from their wiki pages. Also use when the user wants to explore connections between topics in their wiki. Works from any project. Includes an index-only fast mode triggered by "quick answer", "just scan", "don't read the pages", "fast lookup" — returns answers from page summaries and frontmatter without reading page bodies.

Scan the Obsidian wiki and automatically discover missing cross-references between pages. Use this skill when the user says "link my pages", "find missing links", "cross-reference", "connect my wiki", "add wikilinks", "what pages should be linked", or after any large ingestion to ensure new pages are woven into the existing knowledge graph. Also trigger when the user mentions "orphan pages" in the context of wanting to connect them, or says things like "my wiki feels disconnected" or "pages aren't linked well". This is a write-heavy skill — it actually modifies pages to add links, unlike wiki-lint which just reports issues.

Ingest any raw text data, conversation logs, chat exports, or unstructured documents into the Obsidian wiki. Use this skill when the user wants to process data that isn't standard documents or Claude history — things like ChatGPT exports, Slack threads, Discord logs, meeting transcripts, journal entries, CSV data, browser bookmarks, email archives, or any raw text dump. Triggers on "ingest this data", "process these logs", "add this export to the wiki", "import my chat history from X". This is the catch-all for any text source not covered by the more specific ingest skills.

Fetch a URL and distill its content into the Obsidian wiki. If invoked from inside a project directory, the page lands directly in that project's folder (creating the project in the vault if needed). Otherwise it goes to misc/ and gains project affinity over time. Use this skill when the user says "/ingest-url <url>", "add this URL to the wiki", "ingest this link", "save this page", or pastes a URL and says "add this" or "save this to my wiki".

The foundational knowledge distillation pattern for building and maintaining an AI-powered Obsidian wiki. Based on Andrej Karpathy's LLM Wiki architecture. Use this skill whenever the user wants to understand the wiki pattern, set up a new knowledge base, or needs guidance on the three-layer architecture (raw sources → wiki → schema). Also use when discussing knowledge management strategy, wiki structure decisions, or how to organize distilled knowledge. This is the "theory" skill — other skills handle specific operations (ingesting, querying, linting).

Automates ingestion of documents into the Obsidian wiki (obsidian-wiki) using the wiki-ingest pipeline. Handles deduplication via manifest, frontmatter, and cross-links; triggers on user request within the obsidian-wiki project context.

Save the current conversation as a permanent, structured wiki note. Use this skill when the user says "save this", "/wiki-capture", "capture this", "file this conversation", "preserve this", "add this to my wiki", or wants to turn what was just discussed into lasting knowledge. The skill classifies the content, rewrites it as declarative knowledge (not a chat transcript), and places it in the correct vault category.

Use this skill when building, reviewing, debugging, or extending Colyseus multiplayer games, especially TypeScript or Bun projects. Covers authoritative rooms, schema state sync, message validation, matchmaking, room access recipes, reconnection, timing, testing, load testing, and scaling patterns.

Use this skill when building, reviewing, debugging, or extending Excalibur.js games, especially Bun, Vite, TypeScript, or Colyseus-backed game projects. Covers Excalibur engine setup, scenes, actors, lifecycle hooks, resources and loaders, graphics, audio, input, UI, physics, collisions, triggers, actions, timers, ECS, maps, plugins, browser canvas verification, and repo-specific client/server boundaries.

Generate copyable pursue-goal prompts for building repo-local game sprints end to end. Use when Codex should return a prompt the user will manually paste into pursue goal mode, instructing a later run to follow a docs/sprints sprint file task by task using game-sprint-execution for every task until the game is built, verified, completed, or blocked.

Design, build, or reshape distinctive React and CSS interfaces for this browser multiplayer game. Use when changing the lobby, loader, HUD, scoreboard, controls, notices, banners, buttons, responsive layouts, or other DOM-facing game presentation where visual hierarchy, intentional styling, accessibility, and browser screenshot review matter.

Use this skill when brainstorming, defining, scoping, documenting, or planning a browser multiplayer game before implementation. Covers game vision, design pillars, genre and game type selection, player verbs, core loops, MDA, characters, world, mechanics, multiplayer ownership, Excalibur client architecture, Colyseus server authority, Pixel Lab asset planning, prototype slices, game design documents, GDD-style Markdown anchors, and docs/game-design planning artifacts.

Execute repo-local browser multiplayer game sprint tasks after game-sprint-planning has produced docs/sprints artifacts. Use when Codex needs to pick or implement a docs/sprints game task file, continue the next ready task, update sprint/task status and completion evidence, run Bun verification, or implement planned Colyseus, Excalibur, React HUD, SQLite, or Pixel Lab work from the sprint queue.

Turn repo-local browser multiplayer game design docs into execution-ready sprint and task Markdown artifacts. Use when Codex needs to read docs/game-design/*, compose Colyseus, Excalibur, and Pixel Lab implementation guidance, and create or update docs/sprints game-slug sprint and numbered task files for building a game without implementing code.

Package, publish, update, and verify this Bun + Vite + Colyseus browser multiplayer game on itch.io. Use when Codex needs to help prepare an itch.io HTML5 build, add build:itch or package:itch scripts, build with VITE_GAME_SERVER_URL and Vite --base ./, create the HTML5 ZIP, use butler, configure the itch.io project page, run iframe smoke checks, or troubleshoot itch.io asset paths, WebSocket, wss, indexing, draft, restricted, public, or release checklist issues.

Apply Build Right engineering principles as a planning, implementation, and review standard. Use when Codex is choosing technologies, service boundaries, storage models, architecture boundaries, splitting modules or features, creating public contracts, changing provider adapters, reviewing implementation quality or over-engineering risk, planning tests, handling side effects, errors, observability, security, or deciding whether markdown guidance needs enforceable checks.

Execute one bounded task using baseline evidence, narrow implementation, verification, evidence capture, and tracker updates. Use when the user invokes /build-right-execution, wants to work the next issue, run an evidence-driven implementation loop, verify a task, update task state, or continue execution from a prepared task queue.

Maintain Build Right feature planning after preflight and before execution. Use when the user wants to explore a desired feature, groom or reprioritize backlog/sprint work, run bounded product or technical research, use subagents for planning review, create or split execution-ready tasks, update planning evidence, or hand off a ready task to build-right-execution without implementing product code.

Guide founder/product pre-execution setup before AI implementation. Use when the user invokes /build-right-preflight, wants to bootstrap a blank or existing project, capture founder intent, validate assumptions, create product truth, define MVP scope, create operating docs, create Sprint 0 tasks, or prepare the first executable AI task.

Answer one unresolved logic or UI design question with disposable executable evidence. Use for explicit prototype, spike, proof-of-concept, interaction experiment, state-model experiment, or design-experiment requests where observed behavior can reduce uncertainty before production feature planning.

Maintain a durable multi-session decision map when the destination is meaningful but the implementation path cannot yet fit in one trustworthy planning session. Use only when the user explicitly invokes this skill for a foggy effort that needs research, prototypes, founder decisions, or technical decisions before feature planning.

Turn repo-local browser multiplayer game design docs into execution-ready sprint and task Markdown artifacts. Use when Codex needs to read docs/game-design/*, compose Colyseus, Excalibur, and Pixel Lab implementation guidance, and create or update docs/sprints game-slug sprint and numbered task files for building a game without implementing code.

Package, publish, update, and verify this Bun + Vite + Colyseus browser multiplayer game on itch.io. Use when Codex needs to help prepare an itch.io HTML5 build, add build:itch or package:itch scripts, build with VITE_GAME_SERVER_URL and Vite --base ./, create the HTML5 ZIP, use butler, configure the itch.io project page, run iframe smoke checks, or troubleshoot itch.io asset paths, WebSocket, wss, indexing, draft, restricted, public, or release checklist issues.

Use this skill when building, reviewing, debugging, or integrating Pixel Lab API workflows for game-ready pixel art assets, especially in Bun or TypeScript game projects. Covers Pixellab characters, rotations, animations, tilesets, map objects, UI assets, image conversion, async jobs, credential-safe token handling, complete asset-batch manifests, output provenance, and engine integration review.

Prepare, configure, deploy, and verify this Bun + Colyseus + Vite multiplayer game on Railway. Use when Codex needs to help with Railway deployment for this repo, including Dockerfile and .dockerignore setup, Railway game-server and web service settings, DB_FILE_NAME volume persistence, VITE_GAME_SERVER_URL wiring, Railway CLI commands, health checks, logs, smoke tests, or single-replica Colyseus deployment constraints.

Use this skill when brainstorming, defining, scoping, documenting, or planning a browser multiplayer game before implementation. Covers game vision, design pillars, genre and game type selection, player verbs, core loops, MDA, characters, world, mechanics, multiplayer ownership, Excalibur client architecture, Colyseus server authority, Pixel Lab asset planning, prototype slices, game design documents, GDD-style Markdown anchors, and docs/game-design planning artifacts.

Scaffolds the full .claude/ agentic engineering stack for a TypeScript/monorepo project. Analyzes the codebase, then generates CLAUDE.md, rules, agents, hooks, and permissions.

Runs 3 competing implementations of a feature using different strategies, then picks the best one. Swiss cheese layer 1 — diversity of approaches.

Route GPT-5.6 engineering work across eight roles with bounded topologies.

Install, verify, or uninstall the eight schema-7 GPT-5.6 roles and report scoped v0.6 installation health.

Coordinate through HA2HA v1 workspaces using portable protocol files, local folders, or conformant HTTP implementations. Use when Codex needs to publish or join a HA2HA workspace, validate workspace state, update a file with baseVersion, claim a task, add evidence, record a decision, write a handoff, or stop safely on HA2HA conflicts without using MDSync product-only dashboards, comments, admin, stats, hosted auth, or provider sync.

Use hosted MDSync product routes through @mdsync/client while preserving HA2HA protocol boundaries. Use when Codex needs to publish or join a hosted MDSync workspace, inspect dashboard/raw/API state, update files with baseVersion, inspect comments/history/admin/capability state, rotate or verify product tokens, link users to dashboards, write evidence, and stop safely on version_conflict without relying on repo-local scripts.

Coordinate through HA2HA v1 workspaces using portable protocol files, local folders, or conformant HTTP implementations. Use when Codex needs to publish or join a HA2HA workspace, validate workspace state, update a file with baseVersion, claim a task, add evidence, record a decision, write a handoff, or stop safely on HA2HA conflicts without using MDSync product-only dashboards, comments, admin, stats, hosted auth, or provider sync.

Use hosted MDSync product routes through @mdsync/client while preserving HA2HA protocol boundaries. Use when Codex needs to publish or join a hosted MDSync workspace, inspect dashboard/raw/API state, update files with baseVersion, inspect comments/history/admin/capability state, rotate or verify product tokens, link users to dashboards, write evidence, and stop safely on version_conflict without relying on repo-local scripts.