AI & ML

silviobeer Bundle

Execute implementation plans user-story by user-story using agent teams, TDD, wave-scoped Ralph verification, wave gates, and an integration-focused Quality Gate before mandatory QA. Use when: (1) an implementation plan exists and is ready for execution, (2) feature tasks need to be implemented with TDD. Not for: planning, architecture, or requirements.

silviobeer Skill

Turn an approved product vision into a running, empty project: decide the stack and record it as the single source of truth in docs/ARCHITECTURE.md § Stack, run the real scaffold command, verify build and tests are green, and write root AGENTS.md plus the CLAUDE.md pointer. Use when: (1) the product vision exists and the workspace has no application code yet, (2) an existing repo needs its stack recorded and its agent files created. Not for: an existing codebase with code to extract (use intake), product scope (use product-vision), PROJ-level tech design (use architecture), per-PROJ P0 setup (use setup).

sodam-ai Skill

15년+ 디지털 콘텐츠·CG·자동화 경험 기반 생성형 AI 로컬 워크플로우 엔지니어 관점. ComfyUI 같은 로컬 도구의 설치, 워크플로우, 노드, 모델, CUDA·VRAM, API, 업데이트와 복구 작업에서 활성화한다.

ccai40359-wq Bundle

Dual-channel document reading — the main agent reads, one subagent per file reads independently, then cross-compare (match/addition/conflict). MUST USE for close reading, analysis, or fact extraction from documents/PDFs/contracts/manuals/reports/long files — triggers include "read carefully", "don't miss anything", "cross-check", "用子agent读", "双读". Essential when extracting key facts (numbers, limits, clauses, rules). Not for browsing a table of contents, reading a code diff, or merely confirming a file exists.

rospocroccante Skill

Use when spawning subagents or splitting work across models, when an adversarial verdict on code or candidates is needed, or when the user names the setup (Fable as judge, Opus as executor, "model roles", "fable giudice", "opus esecutore", "solo opus", "senza fable"). Encodes the standing division of labour in both variants, with and without a Fable-capable account.

silviobeer Skill

Turn an approved product vision into a running, empty project: decide the stack and record it as the single source of truth in docs/ARCHITECTURE.md § Stack, run the real scaffold command, verify build and tests are green, and write root AGENTS.md plus the CLAUDE.md pointer. Use when: (1) the product vision exists and the workspace has no application code yet, (2) an existing repo needs its stack recorded and its agent files created. Not for: an existing codebase with code to extract (use intake), product scope (use product-vision), PROJ-level tech design (use architecture), per-PROJ P0 setup (use setup).

silviobeer Bundle

Create and curate human-readable project documentation after QA passes. Produces README.md, docs/PROJECT.md, docs/TECHNICAL.md, conditionally merges approved AGENTS.md candidates, keeps CLAUDE.md pointer-only when present, and in framework runs curates the long-lived docs/ baseline behind the P7 gates (curation-caps.sh for form, cross-review for truth). Use when: (1) a feature has passed QA, (2) the user asks for app documentation, feature documentation, technical documentation, diagrams, data model notes, or data-flow explanation, (3) before handoff. Not for: writing PRDs, API reference generation, or first-run onboarding flows.

rospocroccante Skill

Produce a functional design and/or technical design document for a feature, system, or change. Use when the user asks to draft, scaffold, or structure a design doc, spec, RFC, or architecture write-up covering user-facing behavior, requirements, components, data model, APIs, or rollout.

publicdotcom Bundle

Interact with your Public.com brokerage account using the Public.com API. Able to view portfolio, get stock quotes, place trades, and get account updates. To create a Public.com account head to public.com/signup.

silviobeer Bundle

Create and curate human-readable project documentation after QA passes. Produces README.md, docs/PROJECT.md, docs/TECHNICAL.md, conditionally merges approved AGENTS.md candidates, keeps CLAUDE.md pointer-only, and in framework runs curates the long-lived docs/ baseline behind the P7 gates (curation-caps.sh for form, cross-review for truth). Use when: (1) a feature has passed QA, (2) the user asks for app documentation, feature documentation, technical documentation, diagrams, data model notes, or data-flow explanation, (3) before handoff. Not for: writing PRDs, API reference generation, or first-run onboarding flows.

silviobeer Skill

Lean Step 5 variant for frontier models (Claude Fable/Mythos 5.x, Opus 5, GPT-5.x or later): same state.sh, wave-gate, ledger, Quality Gate and QA-handoff contracts as 5_executing, with the step-by-step TDD choreography and persona reviews removed in favor of intent and invariants; the strongest model leads, workers stay tiered by Complexity, reviews go to the strongest opposite model. Use when: (1) wave plans exist and CP1 is sealed, (2) the writer lane runs on a frontier model, (3) a run should be compared against the full 5_executing procedure. Not for: weaker models, planning, architecture, requirements, or a repo without 5_executing installed.

sprintberlin Bundle

Zoho Books MCP endpoint setup, multi-account routing, organization selection, action profiles, helper CLIs, and safe accounting workflows.

hagestedt Skill

Use when measuring cost vs quality by re-running a frozen past-work pack across models and effort/thinking levels on any LLM provider. Provider-agnostic harness; no vendor CLI required.

hagestedt Skill

Use when benchmarking model×effort on Anthropic Claude Platform/Claude Code and you want /claude-api prompt-audit, cost-optimize, and hillclimb on top of the generic frozen-pack harness.

franky47 Bundle

Guide for creating custom OpenCode agents using markdown files with YAML frontmatter

asterxsk Bundle

Computer use with inbuilt messages. Drive the desktop via the clank CLI (any harness, Windows/macOS/Linux). Background control plus sharp blue cursor, click tilt + ripple, typewriter reveal, and message pill with agent-chosen duration.

lathe-cli Skill

Initialize CLI-first applications and keep their OpenAPI contract, generated CLI, and Agent Skill synchronized. Use when creating an application with lathe init or changing a Lathe-generated application.

franky47 Skill

Schedule a prompt to fire back at yourself at a future time or on a recurring cron pattern. Use when the user says "remind me", "every N minutes/hours/days", "at HH:MM tomorrow", "check X periodically", or otherwise asks for any time-based action. Triggered by the `schedule` tool plus the `/schedule` command for listing and deletion.

specsnl Skill

Use this skill whenever the user asks to read an issue and implement it — e.g. "read issue X (github or linear) and implement it", "implement this issue", "pick up GH-123 / ABC-456 and build it". Fetch the issue (GitHub via a GitHub MCP or the gh CLI, or Linear via the Linear MCP), create a new branch, implement the change committing incrementally with clear messages, and STOP before opening a PR. Only create a PR when the user explicitly asks — then defer to the create-pr skill.

specsnl Skill

Use this skill to write, expand, or structure Playwright end-to-end tests for a Laravel + Filament (Spatie-permission) application, especially multi-tenant apps with several panels/roles. Triggers when the user says "add e2e tests", "cover all roles", "write Playwright tests for the panels", "test each role", "e2e coverage per role", or asks to set up Playwright against a Filament admin/tenant panel. It covers the per-role storageState auth model, shared helpers, a per-role coverage matrix (CRUD / state-machine workflows / permission boundaries / tenant isolation), a deterministic test-data strategy, the CI job recipe, and Filament-specific selector conventions. Use it before scaffolding e2e/ so the suite stays maintainable and the role/permission boundaries are actually asserted.

thatmike1 Bundle

Hand a question to ChatGPT's Pro model or its image generator through the user's own chat window, off the Codex meter — the agent packs the prompt and context, the user pastes it and copies the answer back. Use when a plan or design wants a second opinion from another model family, for a large divergent idea run, for self-contained 3D or math-heavy code, for an image, or when the user says gpt-pro or asks to put something to ChatGPT.

aidenxu-1 Bundle

Build or maintain project multi-agent collaboration, including team setup, roles, session routing, handoff and protocol repair. Existing departments handle routine development and review through project role and TASK files.

ad-rian-studio Bundle

Researches the codebase, aligns with the user on requirements, and produces a structured, actionable implementation plan saved to .agents/agent-plans/ before any code is written. Use when asked to plan, outline, design, or architect a feature, refactor, or multi-step task.

ad-rian-studio Bundle

Expert guidelines for auditing, creating, and updating high-signal AGENTS.md files, repository instructions, and agent rules. Use when asked to generate or update AGENTS.md, audit repo instructions, establish durable project guidelines for AI coding agents, or configure repository guidance for OpenCode, Claude Code, Cursor, and Codex.

sayamsingla2000 Skill

Analyze raw prompts, identify intent and gaps, match ECC components (skills/commands/agents/hooks), and output a ready-to-paste optimized prompt. Advisory role only — never executes the task itself. TRIGGER when: user says "optimize prompt", "improve my prompt", "how to write a prompt for", "help me prompt", "rewrite this prompt", or explicitly asks to enhance prompt quality. Also triggers on Chinese equivalents: "优化prompt", "改进prompt", "怎么写prompt", "帮我优化这个指令". DO NOT TRIGGER when: user wants the task executed directly, or says "just do it" / "直接做". DO NOT TRIGGER when user says "优化代码", "优化性能", "optimize performance", "optimize this code" — those are refactoring/performance tasks, not prompt optimization.

sayamsingla2000 Bundle

Tune a user's own LLM prompt when its output doesn't match the required result — drive revisions from observed failures and feedback instead of rewording blind. Works for any prompt type: classification/LLM-judge, structured extraction, query or code generation, summarization, free-text generation. Covers a quick manual pass (a few bad outputs, revise now) and a full automated loop (eval set, held-out scoring, early stopping, multiple starts), and keeps a per-prompt record so a session months later resumes instead of restarting. Use when a prompt isn't producing the needed result, when wrong-vs-expected outputs are available for it, or when a prompt needs systematic improvement against an eval set.

dely0 Bundle

用「干净上下文的第二个 agent」审查代码改动、方案或数据结论：委派一个 spawn 子 agent（不带本会话上下文），强制每条发现带「位置 + 复现命令 + 实测输出」，并用反向验证（把 bug 装回去必须变红）收口。当用户说「审一下 / review 一下 / 找找 bug / 复核这个改动 / 让别的 agent 看看 / 这段代码有没有问题」，或你刚完成会碰数据、凭据、权限、定时任务的改动想验证时使用。

toby-bridges Bundle

OpenClaw skill for local API Relay Audit. Use when an OpenClaw agent must audit a third-party AI API relay, LLM proxy, gateway, or resale API before trusting coding, tool, production, or wallet-sensitive traffic.

toby-bridges Skill

Use when auditing third-party AI API relays, LLM proxies, gateways, or API-key resale services locally before trusting coding, tool, production, or wallet-sensitive traffic.

jadiouo Bundle

Early-Stage Startup Investment Trends (2024–2026)

revenuehunt Bundle

Create a narrated screen-capture demo of a web UI with local voiceover and subtitles. Use when the user asks for a video, screencast, narrated walkthrough, feature demo, or bug explanation. Works with any browser recorder and coding agent with file and shell access.

sukris Bundle

Recall and store durable user preferences, project decisions, recurring facts, and lessons through the configured MemOS MCP tools. Use when prior context may affect a task, when the user asks what was remembered, or when the user explicitly asks to remember or forget something. Do not use for transient chat, raw logs, credentials, or speculative conclusions.

sukris Bundle

Manage noisy build, compile, lint, and test output without losing evidence. Use when Node.js, Java, or other development commands may emit large logs that would flood the model context; skip it for commands whose complete output is already short and clear.

bosmark Bundle

Guide an executive through refining their BoS OS, deepening strategy documents, mapping the roles and responsibilities of people in the organisation, and defining their first mission with phased validation. Runs as an interactive workshop with a challenger-advisor tone. Use when someone wants to improve agent specs, refine strategy documents, create a mission, define decision boundaries, or build an evaluation framework. Triggers: "help me refine my strategy", "my agent spec needs work", "what should my agent's first mission be", "how do I test my agent", "walk me through the workshop", "I've bootstrapped my BoS OS, now what", "executive workshop", "strategy deep-dive", "agent refinement", "mission planning", "ground truth", "calibration". MANDATORY TRIGGERS: BoS OS, workshop, refine, strategy, agent spec, mission, evaluation framework, ground truth, calibrate, decision boundary.

bosmark Bundle

Build the BoS OS for a company, a structured operating system that helps a CEO or founder make better decisions and helps their team understand what they are trying to do. Researches the company from public information, creates the folder structure, populates strategy documents with company-specific content, maps people and their roles, and generates a CLAUDE.md. Triggers: "run the BoS OS", "BoS OS", "Company OS", "bootstrap my company OS", "create an operating system for [company]", "help me organise my AI strategy", "build an agent framework", "I want to do what that talk described", "operating model", "folder structure for agents", "bootstrap my company". MANDATORY TRIGGERS: BoS OS, Company OS, bootstrap, company setup, agent framework, AI strategy, operating model, folder structure.

bosmark Skill

Interview-driven agent spec builder. Takes ONE named agent and conducts a structured interview to produce a finished, runnable spec — covering role, type, decision boundary, escalation triggers, and evaluation criteria. Use when someone wants to spec an agent, define an agent's decision boundary, write an agent spec, build an agent specification, or turn a vague agent idea into a precise runnable document. Triggers: "spec this agent", "help me define my agent", "write an agent spec", "what should my agent be allowed to do", "decision boundary", "escalation triggers", "agent specification". MANDATORY TRIGGERS: agent spec, agent specification, decision boundary, spec builder, spec an agent.