AI & ML

jtaroreh Bundle

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart". Multiple LLM reviewers challenge changes from independent angles.

jtaroreh Bundle

poteto's agent style for concise, detailed responses, deliberate subagents, unslopped prose, simple code, and verified work. Use for poteto, /poteto-mode, or requests to work in this style.

xxjrq Bundle

自动完成 iOS 应用的 App Store 上架工作。Use when an agent needs to inspect an iOS project, build or locate an IPA, upload a build and store assets, fill App Store Connect metadata and compliance information, submit a version for review, check review or publication status, or read a rejection and fix the app or listing. Requires Easy WebBridge and an already logged-in App Store Connect browser.

urmzd Skill

How and when to use the agentspec CLI to manage skills, agents, MCP servers, sessions, memories, and project config across AI coding tools. Use when asked to install or share a skill, register an MCP server everywhere, find a past AI session, or sync agent resources between Claude Code, Codex, Copilot, Gemini, Cursor, and friends.

presidenteog Skill

Hand an image prompt to the user as copy-pasteable text instead of sending it through the Antigravity bridge — for pasting into gemini.google.com (Gemini 3.1 Pro image generation), ChatGPT, Midjourney, or any web image tool. Use this when the user says to give them the prompt, wants to run it themselves, mentions the Gemini website or another web UI, when the Antigravity image quota is exhausted, or when they want to see the prompt text before it gets generated. Outputs the prompt in a copy-ready block plus a manifest of which reference images to attach and which docs the prompt draws on.

presidenteog Skill

Use for mechanical, non-interactive work that can run on a local model instead of burning subscription tokens — bulk classification, mechanical code edits, screenshot/UI description, long-text summarization. Not for anything needing MCP servers, subagents, vault memory, or judgment calls; the local tier has none of that and is slow (seconds to minutes per call, not real-time).

presidenteog Skill

Use for self-contained work that needs real reasoning capability but none of Claude Code's own tooling — research, long summaries, first drafts — where routing to Antigravity's free Gemini/Claude quota via `agy -p` costs nothing instead of burning subscription tokens. Not for anything needing MCP servers, subagents, or vault memory; agy has none of that context.

presidenteog Skill

Build and maintain a per-repo CONTEXT.md, a short prose dictionary of the project's domain terms the agent reads to decode jargon and name things consistently. Use when project terminology is being discussed or disputed, when naming feels inconsistent, when a hard-to-explain concept keeps needing a paragraph, or when writing or editing a CONTEXT.md or an ADR.

farzammohammadi Bundle

Wrap up a session by capturing context, decisions, and state into project-native documentation, then generate a continuation prompt for the next chat. Use when ending a session, switching context, or preparing a handoff.

presidenteog Skill

Use when asked to audit a codebase or module for technical debt and produce a findings report — architectural decay, consistency drift, type/test/dependency gaps, security hygiene. Read-only, cites file:line for every finding, ranks by severity against effort, and never recommends a rewrite. Not for making the fix — pair with a builder agent for that.

presidenteog Skill

Use when the user invokes /better-prompt-text or wants a rough prompt (Spanish or English) turned into a polished, precise English prompt to copy and use elsewhere — text delivery only, no execution, no files.

presidenteog Skill

Use when asked to design an evaluation for an AI agent or skill's real-world reliability — across repeated runs, under noisy tools and environments, against realistic tasks — rather than to verify a single change worked once. Distinct from a one-off QA pass; produces a repeatable eval suite with metrics, noise controls, and human-review gates for high-risk cases.

presidenteog Bundle

Load glTF/GLB models in three.js with GLTFLoader and play their skinned animations with AnimationMixer, including DRACO/Meshopt-compressed meshes and KTX2 textures. Use when importing 3D models into three.js — when the user mentions glTF, GLB, GLTFLoader, AnimationMixer, animation clips, DRACOLoader, or "load a 3D model". For scene/camera/renderer setup use threejs-scene-setup; for materials and lights use threejs-materials-lighting.

presidenteog Skill

Use when deciding how to split a task across the crew roster — whether it needs one agent, several running in parallel, a sequential handoff, or a crew-lead-planned hierarchy — before dispatching, especially when a task looks like it could go to more than one crew-* agent.

presidenteog Skill

Use when a subagent, delegated task, or earlier session's write-up states a count, a file's contents, how a tool behaves, or that something was committed or verified — before acting on it or repeating it to the user.

presidenteog Skill

Use when deciding which model tier a task should run on, investigating why a session or subagent burned more tokens than expected, or doing a periodic check that spend hasn't quietly drifted — grounded in what's actually observable here (session transcripts, /caveman-stats, Model Policy tiers), not a hypothetical enterprise dashboard.

presidenteog Skill

Use when designing or reviewing an agent or skill system and deciding what should live in the prompt versus be pushed out to durable memory, a reusable skill, a typed protocol, or the harness (hooks/permissions) — especially before adding "just one more instruction" to an already-long system prompt or CLAUDE.md.

presidenteog Bundle

Light and shade a three.js scene: choose materials (MeshStandardMaterial PBR vs unlit MeshBasicMaterial), add ambient/hemisphere/directional/point/spot lights, turn on shadow maps, and use an environment map (IBL) for realistic reflections. Use when a three.js model looks black, flat, or wrong — when the user mentions three.js materials, MeshStandardMaterial, lights, shadows, envMap, or PBR. For renderer/loop setup use threejs-scene-setup; for loading models use threejs-gltf-loading.

nerdstalker Bundle

Advise on and critique UI/UX design work through the lens of Luke Wroblewski's published principles, using his MCP archive to retrieve relevant writings, presentations, videos, and other source material for each task. Use this skill whenever the user asks for a design review, UX critique, or feedback on a screen, page, form, sign-up flow, checkout, mobile layout, navigation scheme, or AI/agent interface — or asks "what would LukeW say," mentions Luke Wroblewski, or wants design advice grounded in established best practices rather than taste. Also use it when the user asks to fix or improve a UI and wants the changes justified by recognized principles.

stefancgillberg Bundle

Merge-request / PR review workflow: resolve a branch or MR number against an integration branch (default develop), build a unified diff, and produce a structured review using the session default model and the bundled prompt template; optional Ollama via ./review.sh when present. Use when the user says merge request check, MR review, /merge-request-check, review.sh, or wants a diff-based PR review before merge.

stefancgillberg Skill

Guided MCP server registration. Collects intent conversationally, resolves package details, and delegates to `rai mcp install` or `rai mcp scaffold`. Human never constructs CLI commands.

aaryan1524 Bundle

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

aaryan1524 Skill

Run large codebase migrations and multi-file refactors. Uses the Composio CLI to coordinate issue tracking, batched PRs, and CI verification while the agent executes the transforms locally across hundreds of files.

stefancgillberg Skill

Safe MCP server removal with adapter dependency checking. Shows registered servers, warns about references, deletes config.

stefancgillberg Skill

Health overview of all registered MCP servers plus stack-aware recommendations for missing capabilities. Lists servers, runs health checks, detects project stack, and suggests relevant servers from the governance catalog.

stefancgillberg Bundle

Merge-request / PR review workflow for Python projects (Humanit house ruleset: Clean Architecture, DDD, TDD, CI gates). Resolve a branch or MR number against an integration branch (default main), build a unified diff, run the branch's tests + ruff, and produce a structured review using the session default model and the bundled Python prompt template. Use when the user says merge request check, MR review, /merge-request-check, or wants a diff-based PR review before merge on a Python repo. For iOS/Swift repos use merge-request-check instead.

aaryan1524 Skill

Debug LangChain and LangGraph agents by fetching execution traces from LangSmith Studio. Use when debugging agent behavior, investigating errors, analyzing tool calls, checking memory operations, or examining agent performance. Automatically fetches recent traces and analyzes execution patterns. Requires langsmith-fetch CLI installed.

aaryan1524 Skill

Audit Claude Code to Codex harness migrations, including AGENTS.md/CLAUDE.md scope, hooks, MCP config, skills, secrets, and validation notes. Use after a Codex import or Bring Your AI migration before letting Codex edit code.

aaryan1524 Bundle

Help address review/issue comments on the open GitHub PR for the current branch using gh CLI; verify gh auth first and prompt the user to authenticate if not logged in.

uchimata2 Bundle

Transfers working context between AI sessions — any work, not just coding — so a later session, another agent, or another person can continue seamlessly. Creating or closing a handoff also reconciles your durable homes — a backward sweep that fixes stale tracker statuses, project docs, memory, and index lines the session made out of date — and that sweep can be run on its own, mid-session, without writing a handoff. Use it when wrapping up or pausing significant work, switching agents or sessions, before a context compaction, when a handoff file exists, or when the user says handoff, hand off, resume, continue later, pick up where we left off, take over, save state, reconcile, sweep for stale statuses, tidy up stale statuses, make the tracker match reality, check or validate the handoff config, close out, or wrap up — even if they don't name the skill explicitly.

josefheld Skill

Run any question, idea, or decision through a council of 5 AI advisors who independently analyze it, peer-review each other anonymously, and synthesize a final verdict. Based on Karpathy's LLM Council methodology. MANDATORY TRIGGERS: 'council this', 'run the council', 'war room this', 'pressure-test this', 'stress-test this', 'debate this'. STRONG TRIGGERS (use when combined with a real decision or tradeoff): 'should I X or Y', 'which option', 'what would you do', 'is this the right move', 'validate this', 'get multiple perspectives', 'I can't decide', 'I'm torn between'. Do NOT trigger on simple yes/no questions, factual lookups, or casual 'should I' without a meaningful tradeoff (e.g. 'should I use markdown' is not a council question). DO trigger when the user presents a genuine decision with stakes, multiple options, and context that suggests they want it pressure-tested from multiple angles.

stefancgillberg Skill

Compare knowledge graph against module architecture docs and update drifted fields. Deterministic frontmatter comparison using existing rai graph commands, with inference for narrative sections. HITL before any writes.

stefancgillberg Skill

Close a working session by reflecting on outcomes and feeding structured data to CLI. CLI does all writes atomically; skill does inference reflection.

stefancgillberg Skill

Begin a session by loading context bundle, interpreting it, and proposing work. CLI does all data plumbing; skill does inference interpretation.

kattakath Skill

Do retrieval-augmented generation over the local pgvector store. Use when the user wants to ingest documents/text into a searchable knowledge base, or ask questions answered from that stored corpus ("index this", "add to my notes DB", "what do my docs say about X", "search my knowledge base", "RAG over these files"). Runs entirely local via the `postgres` MCP server + in-DB embeddings (Ollama); no API keys, nothing leaves the machine.

kattakath Skill

Operate a PHYSICAL Android device over ADB from this Mac — wired or wireless — using the fleet `android-phone` CLI (packages/android-phone.nix): pair, connect, disconnect, bootstrap USB→TCP/IP, and mirror via scrcpy. Use when the user asks to "connect my phone", "pair my android", "adb over wifi", "wireless debugging", "mirror my phone screen", "scrcpy", "adb can't see my device", or any ADB connectivity/diagnosis task. NOT for the virtual emulator (that is `android-emu` / the Android Emulator app) and NOT for driving app UIs (that is the gateway's `mobile-mcp` server — this skill gets the device connected so mobile-mcp can).