AI & ML

juliajakubowskabiznes Bundle

Puts work on trial before you ship it. A second AI model (OpenAI Codex, running locally) acts as prosecutor and attacks the work; Claude defends every charge one by one with evidence; a fresh, blind judge rules on each dispute — FIX, DISMISS, or HUMAN DECISION; Claude then repairs only what was ordered. Works on code and on non-code alike: offers, landing pages, plans, documents, spreadsheets. Use whenever the user wants a second opinion, a critique, a review, a sanity check, a red-team pass, or asks "is this actually good", "what did I miss", "poke holes in this", "put this on trial", "tear this apart", "before I send this", or invokes /tribunal. Reach for it especially right after Claude has built or written something substantial, because the model that built a thing is the worst reviewer of it.

juliajakubowskabiznes Bundle

Convenes a virtual advisory board of 4-12 personas (default 4) to debate a business or life decision across 3 rounds of parallel subagents. Each agent reads their persona profile, gives an unfiltered take, engages directly with other members, then lands a final recommendation. Output: one debate file with framed question, all rounds, and a synthesis that preserves real disagreement. Use whenever the user brings a decision, strategic question, dilemma, or situation they want stress-tested — especially when they say "convene the board", "ask the board", "board session", "stress-test this", or invoke "/advisor-board [question]". First run walks the user through a one-time setup (name, board size, roster, optional context files) and self-configures.

phronesis-io Bundle

Identity and profile lifecycle management for an installed EigenFlux Agent. Covers owner-directed Agent Card and setting changes, periodic profile refresh, credential refresh, historical account recovery, account switching, Dashboard access, and CLI server configuration. Use when an existing Agent needs profile or account maintenance, when access is expired (401), or when the user says "set up my profile", "reconnect to the network", "my token expired", "regenerate the claim link", "switch account", "重新生成认领链接", "我要切换账号", "add a server", or "manage servers". Use when the user requests the Console or dashboard address, homepage, or link. Also use when user context has changed and profile needs a refresh. Do NOT use for first-time installation or onboarding (see install.md and ef-onboarding), feed operations (see ef-broadcast), or messaging (see ef-communication).

phronesis-io Bundle

Consume and publish EigenFlux network content. Use on every heartbeat, for Feed pulls and feedback, influence checks, broadcast publishing or deletion, and Agent Attention decisions. Also use when a conversation surfaces a useful signal, offer, need, project update, or milestone worth broadcasting. Authentication is required. Before Console V2 onboarding completes, only consume the read-only baseline Feed and finish its durable batch. During explicit onboarding or in-place upgrade setup, convert qualified baseline items into Attention Prefill. Do not use for private messages.

phronesis-io Bundle

Private messaging, friend management, and real-time streaming for the EigenFlux agent network. Covers sending and receiving messages, managing conversations, friend requests, blocking, and real-time WebSocket streaming of incoming messages via the CLI. Use on every heartbeat cycle to fetch unread messages and reply where appropriate. Also use when user says "message that agent", "reply to the broadcast", "check my messages", "any new DMs?", "add that agent as a friend", "accept friend request", "block this agent", "who are my friends?", "check pending requests", "start streaming messages", or when a feed item's expected_response matches your user's expertise and you can provide actionable information. Also triggers when the user mentions sending a message to someone by name or identifier, e.g. "send XX a message", "tell XX ...", "DM XX", "message XX", "contact XX", "reach out to XX", "reply to XX", "check my inbox", "any new messages?", "add XX as a friend", "check friend requests", "block XX". This includes

435867311 Bundle

Audit AI agent, Codex skill, plugin, MCP, automation, CLI, or developer-tool repositories for concrete security risks. Use when reviewing instructions or code that may affect file access, shell commands, network/API calls, credentials, prompt injection, tool use, generated code execution, dependency installation, or third-party contributions.

igorwarzocha Bundle

Designs, audits, refactors, and packages reusable agent skills. Use for creating or improving SKILL.md files, trigger descriptions, supporting references/scripts/assets, examples, validation, consolidation, or porting skills between agents. Not for one-off prompt edits, strict project procedures, or passive documentation with no repeatable workflow.

igorwarzocha Bundle

Hardens codebases for agent-native maintainability across languages. Use when asked to audit, score, refactor, or plan cleanup for architecture, godfiles, godfunctions, feature folders, duplication, type/contract safety, traversability, feedback loops, worktrees, or subagent-friendly structure.

pymc-labs Bundle

Load whenever the user is working on code that imports pymc, pytensor, or arviz, or asks about Bayesian modeling, MCMC, priors, posteriors, sampling, or model diagnostics. Covers PyMC 6+, PyTensor 3+, ArviZ 1.1+ (DataTree API), pymc-bart, pymc-extras, nutpie, and JAX/NumPyro backends. Use for building probabilistic models, specifying priors, running MCMC, diagnosing convergence, or comparing models. Triggers include: Bayesian inference, posterior sampling, hierarchical/multilevel models, GLMs, time series, Gaussian processes, HSGP, BART, mixture models, prior/posterior predictive checks, MCMC diagnostics, LOO-CV, model comparison, causal inference with do/observe, and any PyTensor Op or graph work.

pymc-labs Bundle

Load when the user is comparing Bayesian models, computing LOO-CV / ELPD, calling arviz_stats.loo or arviz_stats.compare, doing model stacking/averaging, or computing Bayes factors. Covers the ArviZ 1.1 LOO/ELPD/stacking APIs exclusively (no waic). Triggers include: model comparison, LOO, ELPD, compare, loo_expectations, loo_metrics, loo_r2, Pareto k, stacking, Bayes factor, cross-validation, predictive accuracy, information criterion.

pymc-labs Bundle

Load when the user is choosing priors, running prior predictive checks, calling find_constrained_prior, using PreliZ, or otherwise eliciting domain knowledge into a Bayesian model. Covers weakly informative priors, constrained priors, sensitivity analysis, and elicitation workflows. Triggers include: prior selection, elicitation, find_constrained_prior, PreliZ, prior predictive, expert/informative priors, weakly informative priors, constrained priors.

hathach Bundle

Use when you need authoritative hardware/protocol facts from a primary source rather than model memory — an MCU/peripheral datasheet, reference manual (RM/TRM), errata, pinout, register/bitfield layout, memory map, schematic, or the USB spec — before answering register/electrical/timing/protocol questions from training knowledge or the web; or when the user asks to read/open/look up a manual, datasheet, book, or PDF/EPUB from their Calibre library. Requires a local Calibre library at ~/Documents/calibre-library; no-ops if absent.

majiayu000 Bundle

Create, convert, audit, and package executable AI agent workflow patterns as Codex/Claude-style skills. Use when the user wants an agent workflow library, reusable workflow schema, pattern-to-skill conversion, skill pack design, exporter/validator behavior, curated coding-agent patterns, or wants to avoid prompt-library or hardcoded-script designs.

team-attention Bundle

Systematically QA test any application — web apps, native macOS apps, Electron apps, CLI tools, interactive REPLs, or anything on screen. Three modes: browser (agent-browser/Playwright, fast, DOM-level), computer (MCP computer-use, screenshot + pixel clicks, any app), and cli (tmux, send-keys + capture-pane for interactive terminals). Auto-selects mode or accepts --browser / --computer / --cli override. Use when asked to "qa", "QA", "test this site", "test this app", "find bugs", "test and fix", "fix what's broken", "dogfood", "exploratory test", "bug hunt", "QA this app", "사이트 테스트", "앱 테스트", "브라우저 QA", "화면 보고 테스트해줘", "네이티브 앱 테스트", "screen test". Three tiers: Quick (critical/high only), Standard (+ medium), Exhaustive (+ cosmetic). Produces before/after health scores, fix evidence, and a ship-readiness summary.

synapseradio Bundle

Craft or refactor LLM instructions. Use when authoring or revising the prompt for an Agent or Task tool call, or when the user says "write a prompt", "prompt", "prompt an agent", "refactor this prompt", or any similar request.

synapseradio Bundle

Design, strengthen, or audit an Agent Skill. Use before building any new skill — for a researched design brief that skill-creator can build from — and whenever an existing skill needs judgment or change: it over-triggers, under-triggers, bloats its context, drifts from its sources, or needs a quality check. Triggers: "design a skill", "plan a skill before building it", "audit this skill", "review my SKILL.md", "strengthen this skill", "refactor this skill", "is this skill well built", or any request to evaluate or improve how a skill routes, decides, or spends context. Packaging and eval mechanics stay with skill-creator; use this to decide what to build and to verify fit.

posthog Skill

Django migration patterns and safety workflow for PostHog. Use when creating, adjusting, or reviewing Django/Postgres migrations, including non-blocking index/constraint changes, multi-phase schema changes, data backfills, migration conflict rebasing, and product model moves that require SeparateDatabaseAndState. Also use for any deletion or removal of a model, table, column, product, or app — including deleting migration files or retiring a feature — even when no migration is written.

crimsonblazezero Bundle

Audio and media assets for HyperFrames compositions, produced by one shared audio engine (`scripts/audio.mjs`) — multi-provider TTS (HeyGen / ElevenLabs / Kokoro local), background music + sound effects (HeyGen audio-library retrieval by default, with local Lyria / MusicGen BGM generation and a bundled SFX library as the no-credential fallback), Whisper transcription, background removal, and caption authoring. Use for voiceover / TTS, BGM, SFX / sound effects, transcription, captions / subtitles / lyrics / karaoke / per-word styling, voice + provider selection, and music-mood prompting.

crimsonblazezero Skill

Teaches the AI to design like a high-end agency. Defines the exact fonts, spacing, shadows, card structures, and animations that make a website feel expensive. Blocks all the common defaults that make AI designs look cheap or generic.

synthetic-sciences Bundle

Build complex AI systems with declarative programming, optimize prompts automatically, create modular RAG systems and agents with DSPy - Stanford NLP's framework for systematic LM programming

sukumar-elley Skill

Build natural-language crypto/DeFi agents and EVM MCP plugins (Claude Code, Cursor, Codex, Gemini). Aomi turns prompts into wallet-signed txs on Ethereum, Base, Arbitrum, Optimism, Polygon, Linea — non-custodial, fork-simulated. 40+ apps: Uniswap, Aave, Lido, Morpho, GMX, Hyperliquid, Polymarket.

ananthdakoji2001 Skill

Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria.

bianyanhui Bundle

No silent exits — leave a verifiable handoff before a context-heavy pi session stops. Distilled from kimi-lastcall (github.com/kagamiurayama/kimi-lastcall): the agent writes a 5-section handwritten handoff letter (current state / dead ends / unresolved items / first next action / optional context) as a project file, marks it done with a session-bound marker, and the next session verifies and picks up. Use when the user asks for a handoff, handoff letter, 交接, 交接文档, 会话交接, 把当前进度交接给新会话, wants to switch to a fresh session with heavy context, says the session is nearly exhausted / context is running low, or before ending a session that has meaningful unfinished work. The skill is fail-open: it never traps a session, never blocks a stop, and audit records contain decisions only.

bianyanhui Bundle

Double-check workflow. Use when the user asks to double-check, double check, cross-check, verify, cross-review, 复核, 二次检查, 双重检查, or wants an answer checked by a different model before delivery. The agent (the current/main model) completes the task, then a DIFFERENT model (preferring the latest: MiniMax -> MiniMax-M3, DeepSeek -> deepseek-v4-flash) independently checks the result; on failure the main model revises and the loop repeats (max 3 rounds), and the final reply reports which main model and check model were used.

bytepioneer-ai Bundle

为 codexhost 新增 Harness 插件，或规划、审查、补全现有 Harness Adapter。按当前公共契约实现原生能力，区分插件后端、预装发行和 Desktop 产品接入；不用于单纯添加 Model、Provider 或账号。

dynamods Skill

Author and maintain Dynamo agent skills. Use this skill whenever creating a new skill in .agents/skills/, updating an existing skill, fixing drift between canonical skills and Copilot/Cursor/Claude wrappers, or asking how to write effective skill instructions for the Dynamo repo.

hirofumiko Bundle

Delegate implement / review / plan / investigate / security tasks to Cursor CLI (`agent`) from Claude Code. Auto-orchestrates parallel delegation when multiple independent tasks are detected. Triggers on "cursor", "delegate to cursor", "move to cursor", "cursor-cli", "ask cursor", "cursor orchestrate", or when the user wants to hand off a coding task to Cursor — including Japanese phrasings like 「cursorに委譲」「cursorに渡す」「cursorに調べさせて」「cursorでレビュー」「cursorに聞く」.

hirofumiko Skill

Cross-platform readiness setup for the `cursor` delegation skill. Detects the host OS (WSL / Linux / macOS; native Windows is unsupported → WSL), checks every runtime dependency in one pass without spending Cursor tokens, generates the `~/.claude/settings.json` permission allowlist so read-only delegation runs without a prompt, and — once the verdict is READY — interactively seeds a ready-to-use `~/.cursor.json` (a copy of the shipped defaults) so routing tweaks live outside the plugin and survive marketplace updates. Triggers on "/cursor-setup", "cursor setup", "setup cursor", "cursor 環境構築", "cursorのセットアップ", "cursor doctor", or when the cursor skill fails a preflight check (missing agent/jq/timeout/auth).

eliautobot Skill

Send visible agent-to-agent messages through My Virtual World and review communication history safely.

fazer-ai Bundle

Modo operação do fazer.ai agents: debugar conversas em produção e corrigir comportamentos inesperados do agente. Investiga (conversa no Chatwoot + ExecutionLog/flowlog + traces no Langfuse), reproduz no playground, ajusta o agente (prompt/ferramentas/behavior/KB) e re-valida, com toda mutação aprovada. Use quando uma instância JÁ em produção se comporta de forma inesperada e precisa de diagnóstico/ajuste, não é onboarding (subir do zero) nem desenvolvimento de código.

galaxysf31 Bundle

Configure or diagnose Codex Desktop/CLI with a third-party OpenAI-compatible model provider, including secure API-key storage, Responses API verification, and safe switching back to the official OpenAI provider.

heraldquimpao-creator Skill

Build and review general-purpose applications with the Gemini API. Use for text generation, multimodal input, image generation, SDK setup, model selection, tools, function calling, structured output, production hardening, or migrations from older Gemini integrations.

heraldquimpao-creator Skill

Build and review applications with the Gemini Interactions API. Use for new Gemini model integrations, chat, streaming, server-side conversation state, function calling, structured output, image generation, background execution, Deep Research agents, or coordinated tool workflows.

hokupod Bundle

Use Kyoso when the user explicitly asks for multi-agent plan review, security review, CISA Secure by Design review, diff review, or a second opinion from Codex and Claude. Do not invoke implicitly unless the user clearly requests Kyoso or multi-agent review.

hokupod Bundle

Use Kyoso when the user explicitly asks for multi-agent plan review, security review, CISA Secure by Design review, diff review, or a second opinion from Codex and Claude. Do not invoke implicitly unless the user clearly requests Kyoso or multi-agent review.

hokupod Bundle

Use Kyoso when the user explicitly asks for multi-agent plan review, security review, CISA Secure by Design review, diff review, or a second opinion from Codex and Claude. Do not invoke implicitly unless the user clearly requests Kyoso or multi-agent review.