Agent skills

app-screenshots

Capture and curate screenshots of a GUI desktop app for a README, landing page, or docs.

chatlog

Query your own WeChat (微信) history through the local chatlog HTTP API.

dayflow-sync

Sync the caezium/Dayflow fork with upstream: rebase, audit, build, and stop for a pre-install checkpoint.

install-skill

Install a freshly authored skill into the repo, symlink it into the agent dirs, and commit + push.

nibv0.1.1

Turn an idea or article into original hand-drawn editorial illustrations starring a recurring avatar you own.

prune

Audit a codebase for dead code, duplication, and cruft, then safely remove what no longer earns its place.

stats-sync

Sync the caezium/stats fork with upstream: rebase, audit, test, build, and stop before install.

throughline

Publish an implementation plan as a live HTML URL the user can read on their phone, with a Markdown/JSON handoff.

consult-first

Surface the unsettled product and design decisions and get a ruling before freezing them into a deliverable.

domain-modeling

Pin down a project's ubiquitous language and record architectural decisions.

grill-me

Get relentlessly interviewed about a plan or design until every branch is resolved.

grill-with-docs

The same relentless interview, but it writes the ADRs and glossary as it goes.

grilling

Stress-test a plan before building; triggers on any "grill" phrasing.

loop-me

Design recurring personal or work loops as implementation-ready workflow specs; beta and user-invoked.

prototype

Build disposable runnable prototypes to answer consequential design or technical questions.

throughline

Publish an implementation plan as a live HTML URL the user can read on their phone, with a Markdown/JSON handoff.

to-spec

Turn the current conversation into a spec on the issue tracker — synthesis, no interview.

to-tickets

Break a plan or conversation into tracer-bullet tickets, each declaring its blocking edges.

wayfinder

Navigate a large, foggy effort by resolving decision tickets before pretending the work is implementable.

app-screenshots

Capture and curate screenshots of a GUI desktop app for a README, landing page, or docs.

blast-radius

Prove the runtime safety fact most likely to fail when a change crosses boundaries.

clipsmith

Plan and run a custom video-editing pipeline from a plain-language brief: highlight reels, jump-cuts, captions, beat-synced and vertical edits.

clone-site

Clone any URL into a pixel-faithful Next.js codebase: asset/CSS extraction, parallel section builders, animation porting, fidelity verification.

code-review

Review changes since a fixed point along two axes at once: repo coding standards and the originating spec.

codebase-design

Design deep modules with small interfaces, hidden complexity, and testable boundaries.

create-verification-skill

Build a reusable project-local feature map and live verification recipe.

diagnosing-bugs

Reproduce, minimize, hypothesize, instrument, fix, and regression-test difficult failures.

implement

Build a piece of work from an approved spec or set of tickets.

improve-codebase-architecture

Find deepening opportunities, informed by domain language and ADRs.

interrogate

Run an adversarial multi-agent review panel against an implementation, plan, or claim.

maintain-verification-skill

Update a project's verification skill as its surfaces and workflows drift.

make-interfaces-feel-better

Design engineering principles for polished interfaces (animations, hover states, typography, micro-interactions).

nibv0.1.1

Turn an idea or article into original hand-drawn editorial illustrations starring a recurring avatar you own.

prune

Audit a codebase for dead code, duplication, and cruft, then safely remove what no longer earns its place.

resolving-merge-conflicts

Resolve an active merge or rebase from the intent behind both sides, then finish and verify it.

scaffold-exercises

Create exercise directory structures with sections, problems, solutions, and explainers.

superdesign

Design UI, presentations, and graphics on the Superdesign canvas, including reusable systems and model comparisons.

swarm

Run isolated parallel agents with a coverage matrix, resource budget, output contracts, and a root aggregator.

tdd

Test-driven development with red-green-refactor loop.

tldraw-offline

Drive the offline tldraw canvas app: inspect, edit, arrange, connect, lint, and script `.tldraw` files.

triage

Triage a bug or issue to root cause, then plan a TDD-based fix.

ui-skills-root

Route UI work into the smallest relevant upstream UI-skill context.

agents-sdk

Build stateful agents, workflows, realtime apps, and MCP servers on Cloudflare Workers.

cloudflare

Comprehensive Cloudflare platform guidance with current-reference routing.

cloudflare-email-service

Build transactional sending and inbound routing with Cloudflare Email Service.

durable-objects

Create and review Durable Objects, including RPC, SQLite, alarms, WebSockets, and tests.

sandbox-migrate-to-next

Port a stable Sandbox app to the 1.0 preview with explicit cutover checks.

sandbox-next

Build new Sandbox SDK 1.0-preview apps on the matching `@next` package and image.

sandbox-stable

Build against the current stable Cloudflare Sandbox package without mixing preview APIs.

web-perf

Audit Core Web Vitals and diagnose browser performance with Chrome DevTools.

workers-best-practices

Author and review production Workers code against current platform practices.

wrangler

Use the Wrangler CLI across Workers, storage, AI, queues, workflows, and containers.

app-store-preflight-compliance

Scan an iOS/macOS/tvOS/watchOS/visionOS project for App Store rejection risks before you submit.

swiftui-expert-skill

SwiftUI state and `@Observable` data flow, view invalidation and performance, `ForEach` identity, Liquid Glass adoption, and Instruments `.trace` analysis.

typescript-best-practices

TypeScript-specific modeling, validation, exhaustiveness, testing, and telemetry rules.

vue-best-practices

Current Vue 3 composition, reactivity, component, animation, and performance guidance.

vue-pinia-best-practicesv1.0.0

Current Pinia store and reactivity guidance; replaces `pinia-best-practices`.

awesome-design-md

A library of 74 real-brand DESIGN.md design systems (Stripe, Apple, Vercel, Linear, …) to match a brand's look.

brandkit

Premium brand-kit image generation: logo systems, identity decks, and visual-world boards.

design-taste-frontend

Anti-slop frontend skill for landing pages, portfolios, and redesigns; infers the right design direction and ships non-templated interfaces.

design-taste-frontend-v1

The original v1 taste skill, preserved for backward compatibility.

full-output-enforcement

Override LLM truncation: enforce complete, unabridged code output and ban placeholders.

gpt-taste

Elite UX/UI + GSAP motion skill: AIDA structure, wide editorial type, bento grids, scroll-driven motion.

high-end-visual-design

High-end visual design: the exact fonts, spacing, shadows, and animations that make a site feel expensive.

image-to-code

Image-to-code for Codex: generate the design image, analyze it, then build the site to match.

imagegen-frontend-mobile

Generate premium app-native mobile screen concepts and flows inside phone mockups.

imagegen-frontend-web

Generate premium per-section website design reference images (one image per section).

industrial-brutalist-ui

Raw industrial-brutalist UI: rigid grids, extreme type contrast, analog degradation, declassified-blueprint feel.

minimalist-ui

Clean editorial interfaces: warm monochrome, typographic contrast, flat bento grids, no gradients.

redesign-existing-projects

Upgrade existing sites/apps to premium quality; audit generic AI patterns and fix without breaking functionality.

stitch-design-taste

Generate agent-friendly DESIGN.md files enforcing premium, anti-generic UI standards (for Google Stitch).

burrow-system-tools

Diagnose local macOS health and performance through Burrow's MCP tools.

create-rule

Cursor: create a new project rule.

create-skill

Cursor: scaffold a new skill.

create-subagent

Cursor: scaffold a new subagent.

dayflow-pull

Read Dayflow's local database and compute accurate overlap-merged work hours.

dayflow-sync

Sync the caezium/Dayflow fork with upstream: rebase, audit, build, and stop for a pre-install checkpoint.

defuddle

Extract clean markdown from web pages, stripping clutter and navigation to save tokens.

git-guardrails-claude-code

Set up Claude Code hooks to block dangerous git commands.

install-skill

Install a freshly authored skill into the repo, symlink it into the agent dirs, and commit + push.

migrate-to-skills

Cursor: migrate existing rules into skills.

setup-matt-pocock-skills

One-time repo setup for the engineering skills: issue tracker, triage label vocabulary, domain doc layout.

shell

Cursor: shell command helper.

stats-sync

Sync the caezium/stats fork with upstream: rebase, audit, test, build, and stop before install.

update-cursor-settings

Cursor: modify Cursor settings.

vista-dashboard

Publish live agent status into the local Vista dashboard.

wizard

Generate an interactive shell wizard for setup, credentials, migrations, and other human-only procedures.

anthropic-skill-creator

Create, modify, and benchmark portable or Claude-oriented skills with Anthropic's eval workflow; Codex keeps its own managed `skill-creator`.

chatlog

Query your own WeChat (微信) history through the local chatlog HTTP API.

find-docs

Retrieve authoritative, up-to-date technical documentation for any developer technology.

find-skills

Discover and install agent skills.

json-canvas

Create and edit JSON Canvas (.canvas) files: nodes, edges, groups, mind maps, flowcharts.

obsidian-bases

Create and edit Obsidian Bases (.base) with views, filters, formulas, and summaries.

obsidian-cli

Interact with Obsidian vaults via the CLI; also plugin/theme development and debugging.

obsidian-convert

Convert PDFs/DOCX into Obsidian markdown notes, preserving content verbatim.

obsidian-markdown

Author Obsidian Flavored Markdown: wikilinks, embeds, callouts, properties.

obsidian-vault

Search, create, and manage notes in an Obsidian vault.

teach

Maintain a stateful teaching workspace with short lessons, references, and learning records.

technical-writing

Apply a user-invoked Diátaxis and sentence-clarity pass to technical documents.

vault-resync

Re-pull every external capture source (Feishu, WeChat, Zen tabs, live Claude sessions) into the Obsidian inbox and refresh the Command Center dashboards.

writing-for-agents

Write briefs, pointer docs, acceptance contracts, and context that independent agents can execute reliably.

digest

Aggregate top RSS feeds into a daily tech recap.

how

Explain a subsystem's architecture and runtime flow, with an optional independent critique pass.

kaggle-grind

Run a CV-first, OOF-driven, resume-safe Kaggle competition loop with honest acceptance gates.

last30daysv3.21.1

Deep research engine covering the last 30 days across 10+ sources.

nvidia-kaggle-skill

Kaggle competition research and submission workflow: overview fetches, writeups, discussion/kernel mining, and dataset uploads.

research

Produce durable, cited technical-research artifacts from primary sources.

why

Reconstruct the evidence behind a design, threshold, regression, or trade-off across every available source category.

bro

Restate the last message in plain human language, no jargon.

handoff

Compact the current conversation into a handoff document for another agent to pick up.

henry-mode

Route long-running cross-session work through recall, consultation, proof, swarms, evidence trails, and reflection.

i-have-adhd

Lead with the next action, number multi-step work, restate state, suppress tangents. Stays on until "stop adhd mode".

playbook

Pick the right ordered sequence of skills for the job, announce it, and name the next one after each stage.

recall

Reconstruct relevant work across Claude, Codex, backups, and current repository state.

recap

Consolidate everything that changed this session into one authoritative message plus a hand-test checklist.

reflect

Turn surprising or costly work into a reviewed test, script, rule, skill, or architectural improvement.

show-me-your-work

Keep a durable decision, evidence, dead-end, and recovery trail during expensive or autonomous work.

principle-boundary-discipline

Parse and validate at external boundaries, then trust typed internal code.

principle-build-the-lever

Prefer a rerunnable tool or harness over repeated one-off labor.

principle-guard-the-context-window

Delegate exploration and return compact conclusions plus artifact pointers.

principle-make-operations-idempotent

Make retries, restarts, and partial runs converge to the same result.

principle-prove-it-works

Treat runtime evidence, not confidence or proxies, as the completion condition.

principle-separate-before-serializing-shared-state

Give concurrent actors separate owned state before reaching for locks.

principle-sequence-verifiable-units

Decompose work into independently provable units that leave useful state.

principle-type-system-discipline

Make invalid states and unhandled variants fail at compile time.

1-1-prep

Prepare for a 1:1 with context from person file, meeting history, and shared projects.

daily-roundup

Morning briefing combining email, meetings, and Slack.

decision-extractor

Scan recent notes for decisions and offer to update Decision Logs.

friday

End-of-week wrap-up for the weekly report.

monday

Monday morning setup: week folder, carry-forward, briefing.

week-prep

Generate a briefing with calendar, tasks, and attention flags.

baseline-ui

Quickly deslop UI code by fixing spacing, hierarchy, typography, and small layout issues.

buildspace-ci-cd

Configure and troubleshoot BuildSpace reusable GitHub Actions workflows and blocks for automated releases.

chat-adapter-imessage

Connect the Vercel AI SDK to iMessage.

create-adaptable-composable

Create a library-grade Vue composable that accepts maybe-reactive inputs (MaybeRef / MaybeRefOrGetter) so callers can pass a plain value, ref, or getter.

create-design-md

Create or update a DESIGN.md from an existing product repository or public website.

fixing-accessibility

Audit and fix HTML accessibility issues including ARIA labels, keyboard navigation, focus management, color contrast, and form errors.

fixing-metadatav1.0.1

Audit and fix HTML metadata including page titles, meta descriptions, canonical URLs, Open Graph tags, Twitter cards, favicons, JSON-LD structured data, and robots directives.

fixing-motion-performance

Audit and fix animation performance issues including layout thrashing, compositor properties, scroll-linked motion, and blur effects.

imessage

Build, debug, or migrate iMessage integrations with Photon.

improve-ui

Audit an existing product surface against its own design evidence, identify verified UI problems, and write self-contained implementation plans for another agent.

photon-cli

Use when working with the Photon CLI — the `photon` binary (alias `pho`). Reach for this skill to set up / bootstrap Photon for a user and run it yourself non-interactively (create a project, capture the id + secret, verify) — when the user says "set this up for me," that is your job to do, not hand back. Also covers running any CLI command, authenticating (device-authorization login / logout / whoami / auth status), managing projects (create, show, list, rename, regenerate the Spectrum API secret, delete, open in the dashboard), managing Spectrum resources (lines, users, platforms, profile, avatar), resolving config and environment (PHOTON_PROJECT_ID, PHOTON_TOKEN, PHOTON_API_HOST, credentials storage, multi-backend), and billing/upgrades (plans, checkout, manage, `projects upgrade`). Answers operational questions like "how many lines do I have?", "what's my project secret?", "how do I make this a business line?", and "how do I point the CLI at a different backend?". This is the entry point; the happy-path setup is right below — consult the topic files for full reference. Keywords: photon, pho, photon cli, photon-ai cli, @photon-ai/cli, photon dashboard, command line, terminal, login, device authorization, oauth device flow, logout, whoami, auth status, credentials, PHOTON_PROJECT_ID, PHOTON_TOKEN, PHOTON_API_HOST, PHOTON_CONFIG_DIR, projects, create project, bootstrap, project id, project secret, regenerate-secret, rotate secret, check-phone, upgrade, billing, plans, checkout, stripe, manage, portal, free tier, pro, business, enterprise, dedicated line, spectrum, lines, add line, imessage line, users, invite, platforms, enable, disable, profile, avatar, ping, env, config, json output, api-host, multi-backend, backend, bun, commander, eden, dashboard-api, quickstart, agent setup, automated setup, non-interactive, scriptable bootstrap, set up photon.

spectrum

Build or review messaging agents with Photon Spectrum (`spectrum-ts`).

vue-debug-guides

Vue 3 debugging and error handling for runtime errors, warnings, async failures, and SSR/hydration issues.

vue-jsx-best-practicesv2.0.0

JSX syntax in Vue (e.g., class vs className, JSX plugin config).

vue-options-api-best-practicesv2.0.0

Vue 3 Options API style (data(), methods, this context).

vue-router-best-practicesv1.0.0

Vue Router 4 patterns, navigation guards, route params, and route-component lifecycle interactions.

vue-testing-best-practicesv1.0.0

Use for Vue.js testing.