agentsclimarketplace

Context & AI engineering

25,729 rows, by stacks then stars

1,865 of these skill files have been read, by 1,328 distinct authors, and what they tell an agent counted. What Context & AI engineering authors agree on, and what they forbid

  • Nts add mcp

    juncoding/nextjs-trpc-prisma-starter/skills/nts-add-mcp Skill

    0★ repo

    Add an MCP (Model Context Protocol) entry point to an existing project scaffolded with nextjs-trpc-prisma-starter. Use this when the user wants AI clients (Claude Desktop, Cursor) to query the project's data over OAuth, mentions MCP, wants to expose tools to AI, or invokes /nts-add-mcp. Wires up /api/mcp/route.ts, the MCP plugin in Better Auth (the project's OAuth provider), the .well-known OAuth discovery endpoints, a tool registry at src/server/mcp/registry.ts, one example tool, and the migration for the three OAuth tables. Requires Better Auth to be wired — if not, prompts to run /nts-add-auth first.

  • Clean mcp

    AntonioTimo/skillchecker/examples/clean-mcp Skill

    0★ repo

    Looks up a known MCP server's transport and endpoint from a bundled reference catalogue, and explains in prose how the user would add it themselves. Pure read-only data lookup — it never registers anything.

  • Mcp builder

    decebal/curated-claude-skills/skills/stack-agnostic/mcp-builder Skill

    no license0★ repo

    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).

    The same file is in 42 other repositories

  • Skill creator

    decebal/curated-claude-skills/skills/stack-agnostic/skill-creator Skill

    no license0★ repo

    Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

    The same file is in 19 other repositories

  • Datadata memory

    datadata-team/datadata-skills/skills/datadata-memory Skill

    no license0★ repo

    Manage AI persistent memory via the Datadata MCP Server — add, search, update, and delete memories, with semantic search and multi-dimensional filtering. Use when the user wants the AI to remember information, recall previous conversations, or manage persistent knowledge. Triggers: remember, memory, recall, don't forget, note this down, previously mentioned, search memory, delete memory.

  • Find mcp servers

    unitedideas/claude-skills-foundry/skills/find-mcp-servers Skill

    0★ repo

    Use Not Human Search (nothumansearch.ai) to discover MCP servers and agent-friendly sites for a given task. Invoke when the user asks "what MCP server does X", "find me an MCP for Y", "is there an agent endpoint for Z", or otherwise needs to locate a Model Context Protocol server, agent-ready API, or LLM-compatible service.

  • Context engineering

    mustafayilmazart/kesif-claude-skills/skills/context-engineering Skill

    0★ repo

    Context Engineering ve PRP (Plan-Research-Plan) sistemi. Feature geliştirme öncesi kapsamlı bağlam hazırlığı yapar. INITIAL.md ile feature spec, PRP ile implementation blueprint, examples ile pattern referansı sağlar. Büyük özelliklerde ve yeni projelerde kullan.

  • Hybrid memory

    mustafayilmazart/kesif-claude-skills/skills/hybrid-memory Skill

    0★ repo

    LLM Wiki pattern + Cognee tarzı hibrit hafıza — Vector (ChromaDB) + Graph (Neo4j/NetworkX) + Relational (SQLite) üçlüsü. Oturumlar arası kalıcı bellek için kullan; kullanıcı "hatırla", "not al", "önceki sohbet", "ne konuşmuştuk" dediğinde. Retrieve-based RAG'den farklı — periyodik **compile** ederek Obsidian uyumlu wiki üretir.

    The same file is in 1 other repository

  • Skill creator

    mustafayilmazart/kesif-claude-skills/skills/skill-creator Skill

    0★ repo

    Yeni skill oluşturma, mevcut skill'leri geliştirme ve performans ölçümü. Sıfırdan skill yazmak, mevcut skill'i güncellemek/optimize etmek, eval testleri çalıştırmak, varyans analizi ile benchmark almak veya skill description'ını tetikleme doğruluğu için optimize etmek istediğinde kullan. Kullanıcı 'skill oluştur', 'skill yaz', 'skill güncelle', 'skill test et' dediğinde tetikle.

  • Plan capacity

    GRIDLOCK-NYC/claude-skills/skills/plan-capacity Skill

    0★ repo

    Assess whether a task is feasible for a Claude Code agent and produce a dispatch plan. Decomposes tasks into atomic units, classifies each using Anthropic's proven decision ladder, checks delegation specs, and outputs a concrete execution plan. Use when user says 'can an agent handle this', 'is this too much for one agent', 'agent capacity', 'scope this task', 'how should I break this down', 'task scoping', 'will this thrash', 'dispatch plan', 'how many agents', or 'is this a single agent job'. Runs after a plan is produced to validate the plan's task breakdown. Do NOT use for generating the plan itself (use plan mode for that). Do NOT use for reviewing plan quality (use plan-review for that). Do NOT use for choosing between architecture approaches (use plan-directions for that).

  • Skill builder

    GRIDLOCK-NYC/claude-skills/skills/skill-builder Skill

    0★ repo

    Build new Claude Code skills with best-practice enforcement from Anthropic's official skill guide. Handles use case definition, pattern selection, frontmatter generation, instruction writing, validation, and test case generation. Use when user says "build a skill", "create a skill", "make a new skill", "turn this into a skill", "I want a skill for", "skill for [X]", "make me a slash command", or "new skill that [does something]". Do NOT use for running evals or benchmarking existing skills (use skill-eval for that). Do NOT use for listing, deleting, or inspecting existing skills.

  • Skill eval

    GRIDLOCK-NYC/claude-skills/skills/skill-eval Skill

    0★ repo

    Benchmark how closely a skill followed its own instructions. Reads the skill's SKILL.md, extracts verifiable requirements, scores the skill's output from the current session. Returns compliance % with per-requirement pass/fail and cited evidence for every miss. Use when user says 'skill eval', 'did that skill work', 'how well did it follow instructions', 'benchmark skill', 'skill compliance', 'grade that skill', 'skill score', 'evaluate skill', 'skill audit', 'test skill'. Do NOT use for session-level audits (use session-audit for that). Do NOT use for building skills (use skill-builder for that). Do NOT use for behavioral evals of project code (use eval for that).

  • Jc handoff

    jincheng2026/jcskill/skills/jc-handoff Skill

    0★ repo

    Generate a structured handoff package for starting a new Claude Code session. Extracts tasks, progress, decisions, rejected approaches, file changes, and implicit preferences from the current conversation into a copy-paste-ready startup document. Use when user says: /handoff, handoff, session handoff, pack context, migrate session, new session, or in Chinese: 打包上下文, 交接, 迁移会话, 开新会话, 会话交接.

  • Vibe to agentic framework

    shekerkamma/cc-best-practice/.claude/skills/presentation/vibe-to-agentic-framework Skill

    0★ repo

    The conceptual framework behind the presentation — what "Vibe Coding to Agentic Engineering" means, why the journey is structured the way it is, and how every slide fits the narrative arc

    The same file is in 1 other repository

  • Audit skill

    gnomatix/steering/.agent/skills/audit-skill Skill

    no license0★ repo

    Audit a SKILL document — Antigravity .agent/skills/<name>/SKILL.md, Claude Code SKILL.md, Cursor agent skill .mdc, Kiro skill, or equivalent invocable capability with frontmatter. Runs the standard audit checklist plus reference-exemplar comparison plus portability/graceful-degradation audit (cross-platform support, OSS alternatives to proprietary services). Use when reviewing or revising any invocable skill file. SKIP for always-on rules (use audit-rule) or multi-step workflow documents (use audit-workflow).

  • Audit collection

    gnomatix/steering/.claude/skills/audit-collection Skill

    no license0★ repo

    Audit a body of related agent documents — a `.kiro/` tree, a `~/.claude/skills/` directory, a project's `docs/agent/` collection, or any directory of RULE/SKILL/WORKFLOW files. Runs per-document audits plus the cross-document refactoring pass (split / join / merge / clone / reformat). Use when reviewing the structure of a documentation set, not just one file. For single-document audits, use audit-rule, audit-skill, or audit-workflow directly.

  • Audit rule

    gnomatix/steering/.claude/skills/audit-rule Skill

    no license0★ repo

    Audit a behavioral RULE document — CLAUDE.md, steering file, system prompt directive, project convention, .cursorrules, .cursor/rules/*.mdc, .kiro/steering/*.md. Catches behavior-vs-surface failures, embedded biases, escape hatches, and other ways the rule can be technically followed while violating its intent. Use when reviewing or revising any document that prescribes always-on or auto-included agent behavior. SKIP for invocable skill files (use audit-skill) or multi-step workflow documents (use audit-workflow).

  • Audit skill

    gnomatix/steering/.claude/skills/audit-skill Skill

    no license0★ repo

    Audit a SKILL document — Claude Code SKILL.md, Cursor agent skill, Kiro skill, or equivalent invocable capability with frontmatter. Runs the standard audit checklist plus reference-exemplar comparison plus portability/graceful-degradation audit (cross-platform support, OSS alternatives to proprietary services). Use when reviewing or revising any invocable skill file. SKIP for always-on rules (use audit-rule) or multi-step workflow documents (use audit-workflow).

  • Ai memory setup

    fabioc-aloha/Alex_ACT_Edition/.github/skills/ai-memory-setup Skill

    0★ repo

    Detect, resolve, and manage the Alex_ACT_Memory shared memory bus. Fires on bootstrap, session start (announcements), and feedback writes.

  • Anti hallucination

    fabioc-aloha/Alex_ACT_Edition/.github/skills/anti-hallucination Skill

    0★ repo

    Prevent fabricated facts, invented APIs, and citation confabulation at the point of generation. Use when generating factual claims, code examples, API references, library names, configuration values, error messages, or citations — anything where 'sounds plausible' is not the same as 'is real'.

  • Prompt creator

    fabioc-aloha/Alex_ACT_Edition/.github/skills/prompt-creator Skill

    0★ repo

    Create prompts that pass prompt-review's five gates by construction — intent capture, slash-command naming, single-workflow scope, draft as numbered steps, dogfood self-review. Use when authoring a new prompt, refactoring an existing one, or promoting a Mall prompt into the heir's brain.

  • Prompt review

    fabioc-aloha/Alex_ACT_Edition/.github/skills/prompt-review Skill

    0★ repo

    Audits a candidate prompt (.prompt.md) against five gates (spec compliance, content quality, scope fit, safety, currency & coherence). Use when reviewing a new prompt draft before commit, evaluating a Mall prompt or store prompt for adoption, or re-auditing existing prompts on a periodic cadence.

  • Skill creator

    fabioc-aloha/Alex_ACT_Edition/.github/skills/skill-creator Skill

    0★ repo

    Create skills that pass skill-review's five gates by construction — intent capture, prior-art scan, draft against gates, dogfood self-review. Use when authoring a new skill, refactoring an existing one, or adopting a Mall unit into this brain.

  • Nfs add mcp

    juncoding/nextjs-fullstack-starter/skills/nfs-add-mcp Skill

    0★ repo

    Add an MCP (Model Context Protocol) entry point to an existing project scaffolded with nextjs-fullstack-starter. Use when the user wants AI clients (Claude Desktop, Cursor) to query the project's data over OAuth, mentions MCP, wants to expose tools to AI, or invokes /nfs-add-mcp. Wires up /api/mcp/route.ts inside an (mcp) route group, the MCP plugin in Better Auth (the project's OAuth provider), the .well-known OAuth discovery endpoints, a tool registry at src/server/mcp/registry.ts, one example tool, and the migration for the three OAuth tables. Requires Better Auth to be wired — if not, prompts to run /nfs-add-auth first.

  • Socratic product

    Kimotep/skills/socratic-product Skill

    0★ repo

    Use this skill when a user wants to design an agentic system before writing any code. Triggers include: "help me plan an AI workflow", "I want to build an agent system", "help me structure my agentic project", "I have an idea for an AI tool", or any request to think through multi-agent logic, file scaffolding, or system design for an LLM-powered project. This skill runs a structured, time-boxed Socratic interview (target: 30 minutes) using multiple-choice UI for every question, and produces a linked set of MD documents plus a harness-specific root config file ready to drop into the repo — including a starter repo structure tailored to the project's type and stack, agent specs written to a defined quality standard (responsibilities, I/O contracts, tool scope, and failure handling), and a root config file for Claude Code, Cursor, Windsurf, OpenCode, or other AI coding tools.

  • Voice compass

    Kimotep/skills/voice-compass Skill

    0★ repo

    Use this skill when a user wants to build a personalized writing assistant — a copywriter, content writer, or "writes like me" agent — that captures their own voice, tone, and style. Triggers include: "help me build an AI that writes like me", "I want a copywriter for my brand/newsletter/blog", "create a writing assistant for this project", "set up a style guide so Claude writes in my voice", or any request to define personality, voice, tone, or writing rules for an agent producing blog posts, marketing copy, emails, social posts, or messages. Consider this skill whenever the user wants AI-generated content to "sound like them," even without the words "skill" or "agent". Runs a structured Socratic interview (~30 minutes), optionally analyzes writing samples the user provides, and produces a new, installable writer skill ready to drop into any project.

  • Add mcp ollama

    yukihirop/nagi/.claude/skills/add-mcp-ollama Skill

    0★ repo

    Add Ollama MCP plugin for local LLM access from agent containers. Triggers on "add ollama", "setup ollama", "enable ollama".

  • Add mcp vercel

    yukihirop/nagi/.claude/skills/add-mcp-vercel Skill

    0★ repo

    Add Vercel MCP plugin for deploying websites from agent containers. Triggers on "add vercel", "setup vercel", "enable vercel".

  • Change claude code

    yukihirop/nagi/.claude/skills/change-claude-code Skill

    0★ repo

    Switch agent back to Claude Code. Triggers on "change claude code", "switch to claude code", "use claude code", "claude code に切り替え".

  • Create container plugin agent hooks

    yukihirop/nagi/.claude/skills/create-container-plugin-agent-hooks Skill

    0★ repo

    Scaffold a new agent-hooks plugin for nagi containers. Generates index.mjs with hook factories and deploy/templates/container/claude-code/entry.template.ts registration. Triggers on "create agent hooks plugin", "new agent hooks", "scaffold agent hooks".

  • Create container plugin mcp

    yukihirop/nagi/.claude/skills/create-container-plugin-mcp Skill

    0★ repo

    Scaffold a new MCP plugin for nagi agent containers. Generates package, Dockerfile entry, and deploy/templates/host/entry.template.ts registration. Triggers on "create mcp plugin", "new mcp plugin", "add mcp", "scaffold mcp".

  • Create group prompt

    yukihirop/nagi/.claude/skills/create-group-prompt Skill

    0★ repo

    Create prompt files (IDENTITY.md, SOUL.md, INSTRUCTIONS.md etc.) in deploy/{ASSISTANT_NAME}/groups/{channel}/{group}/. Writes directly to the user-editable default layer (skipping templates). Interactive setup for agent personality and behavior. Triggers on "create group prompt", "add group prompt", "create prompt", "add identity", "add soul".

  • Setup opencode

    yukihirop/nagi/.claude/skills/setup-opencode Skill

    0★ repo

    Set up Open Code agent as an alternative to Claude Code. Supports OpenRouter, Gemini, and OpenAI providers. Triggers on "setup opencode", "setup open code", "use opencode", "switch to opencode".

  • Update group prompt

    yukihirop/nagi/.claude/skills/update-group-prompt Skill

    0★ repo

    Interactively update an existing group prompt file (CLAUDE.md, AGENTS.md, IDENTITY.md etc.) under deploy/{ASSISTANT_NAME}/groups/{channel}/{group}/. Asks the user at every step and previews a diff before saving. Triggers on "update group prompt", "edit group prompt", "modify claude.md", "edit identity", "update soul", "update instructions".

  • Update groups

    yukihirop/nagi/.claude/skills/update-groups Skill

    0★ repo

    Sync group defaults (CLAUDE.md etc.) from deploy/{ASSISTANT_NAME}/groups/ to __data/{ASSISTANT_NAME}/groups/. Use after editing group defaults, or when the agent's CLAUDE.md needs refreshing. Triggers on "update groups", "sync groups", "refresh claude.md", "update agent config".

  • Agent browser

    yukihirop/nagi/container/skills/agent-browser Skill

    0★ repo

    Browse the web for any task — research topics, read articles, interact with web apps, fill forms, take screenshots, extract data, and test web pages. Use whenever a browser would be useful, not just when the user explicitly asks.

    The same file is in 2 other repositories

  • Prompt refinement

    marzun9620/agent_skills/meta/skills/prompt-refinement Skill

    no license0★ repo

    Refines a rough instruction into a structured Agent Task Prompt before another agent executes. Use when the user is about to delegate work to a subagent, Codex CLI, Cursor, or another coding agent and wants a clear, complete brief with goal, context, files to inspect, constraints, deliverables, verification, and edge cases. Triggers - "/refine", "refine this prompt", "make this prompt better", "brief an agent", "structure this before running", "improve my instruction".

  • Setup matt pocock skills

    marzun9620/agent_skills/meta/skills/setup-matt-pocock-skills Skill

    no license0★ repo

    Sets up an `## Agent skills` block in AGENTS.md/CLAUDE.md and `docs/agents/` so the engineering skills know this repo's issue tracker (GitHub or local markdown), triage label vocabulary, and domain doc layout. Run before first use of `to-issues`, `to-prd`, `triage`, `diagnose`, `tdd`, `improve-codebase-architecture`, or `zoom-out` — or if those skills appear to be missing context about the issue tracker, triage labels, or domain docs.

    The same file is in 15 other repositories

  • Write a skill

    marzun9620/agent_skills/meta/skills/write-a-skill Skill

    no license0★ repo

    Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.

    The same file is in 20 other repositories

  • Playwright cli

    marzun9620/agent_skills/testing/skills/playwright-cli Skill

    no license0★ repo

    Use when running Playwright via terminal CLI — `npx playwright test` (test runner), `codegen` (interactive recording), `screenshot` / `pdf` (one-off captures), and CI sharding. NOT for agent-driven real-time browser control (use `claude-in-chrome` MCP tools for that).

  • Hv work

    l4ci/hv-skills/hv-work Skill

    0★ repo

    Orchestrator-driven parallel implementation — plans tasks, dispatches worker subagents, verifies, commits atomically per task. Supports branch or worktree isolation and direct merge or PR. Use when items already exist in BACKLOG.md and need implementation ("implement [B07]", "build these"); for an item not yet captured use /hv-go.

  • Llm wiki

    aksheyw/claude-code-learned-skills/skills/llm-wiki Skill

    0★ repo

    Persistent project wiki that compounds knowledge across sessions. Ingest learnings, query past decisions, lint for contradictions. Based on Karpathy's LLMwiki pattern.

  • Prompting fable 5

    aksheyw/claude-code-learned-skills/skills/prompting-fable-5 Skill

    0★ repo

    How to prompt and scaffold Claude Fable 5 / Mythos 5 — effort levels, long-horizon runs, parallel subagents, memory, the send-to-user tool, and how to reduce false-positive safety fallbacks on legitimately benign work (Opus 4.8 is the intended fallback). Use when starting Fable-5 work, migrating skills/prompts from Opus, or diagnosing why benign work keeps falling back to Opus. Note: if a task is genuinely in a restricted domain, the fallback is intended safety behaviour, not something to work around.

  • Prompting opus 4 8

    aksheyw/claude-code-learned-skills/skills/prompting-opus-4-8 Skill

    0★ repo

    How to prompt and scaffold Claude Opus 4.8 — effort/thinking defaults, verbosity, literal instruction-following, tool-use and subagent triggering, code-review recall, and the cream/serif frontend house-style that's wrong for dashboards. Use when starting Opus 4.8 work, migrating prompts from Opus 4.7, tuning verbosity or effort, wondering why the model isn't "thinking", seeing a code-review recall drop, or getting a cream+serif design you didn't ask for.

  • Prompting qwen3 coder

    aksheyw/claude-code-learned-skills/skills/prompting-qwen3-coder Skill

    0★ repo

    How to prompt and run Qwen3-Coder (qwen3-coder:30b via Ollama) — a NON-thinking model with no deliberation channel, optimised as an agentic patch-writer (though it CAN critique — measured). Covers the official sampling params, tool calling as the vendor's structured path, the 256K/YaRN context story, and the "silent or flooding" failure signature — including the measured finding that a reviewer prompt's own suppression clauses, NOT grammar-forced JSON, silence it. Use when prompting Qwen3-Coder, wiring it into a local council/review seat, debugging an empty or fabricating seat, tuning sampling, or deciding whether it is the right model for a job.

  • Prompting sonnet 5

    aksheyw/claude-code-learned-skills/skills/prompting-sonnet-5 Skill

    0★ repo

    How to prompt and scaffold Claude Sonnet 5 — the migration-from-4.6 API breakers (sampling params now error, new tokenizer inflates tokens ~30%, extended-thinking budgets removed), adaptive-thinking-on-by-default, effort mapping, tool-use triggering, and design/code-review tuning. Use when starting Sonnet 5 work, picking Sonnet as a subagent model, migrating a 4.6 prompt or n8n/API call, hitting a 400 error on temperature, or seeing max_tokens truncation.

  • Ship helpbot

    taskmasterpeace/ship-pack/skills/ship-helpbot Skill

    0★ repo

    Builds an embeddable "ask the guide" assistant for any app: chunks the project's user-guide markdown into docs/guide-index.json, emits a framework-agnostic <ship-helpbot> web component, and a reference /api/ask-docs route that retrieves the top sections and asks an LLM to answer WITH citations to guide anchors. Model-pluggable (Claude API, Vercel AI Gateway, or local Ollama). Use when the user wants an in-app help bot, docs chatbot, "ask the docs/guide" widget, support assistant, RAG over the user guide, embeddable help search, or "answer questions from my documentation." Triggers on "ask the guide", "help bot", "docs chatbot", "ask-docs", "in-app help assistant", "chat with my user guide", "guide search widget".

  • Mcp rate limit handler

    sanjibani/mcp-skills-pack/skills/mcp-rate-limit-handler Skill

    0★ repo

    When an MCP tool call is throttled (429), decides whether to wait, batch, parallelize, or surface to the user. Avoids the agent loop hammering a rate-limited endpoint. Use when a tool returns 429, "rate limit hit", or when you've made 3+ rapid calls to the same MCP in a session. Triggers on "rate limit", "429", "throttled", "slow down", "too many requests".

  • Mcp schema discoverer

    sanjibani/mcp-skills-pack/skills/mcp-schema-discoverer Skill

    0★ repo

    Looks up the JSON schema, return type, and required fields of any MCP tool before you call it. Prevents wasted calls on tools whose input shape you guessed wrong. Use when about to call an MCP tool you haven't called in this session, or when the previous call returned a "missing required field" error. Triggers on "what fields does this tool need", "tool schema", "JSON schema for", "input shape", "required params".

  • Mcp tool picker

    sanjibani/mcp-skills-pack/skills/mcp-tool-picker Skill

    0★ repo

    When the user has multiple MCP servers connected, picks the right one for the task and explains why. Avoids wasted tool calls on wrong-server probes. Use when the user asks a question that could be answered by more than one MCP, or when you've already made 2+ tool calls without progress. Triggers on "which MCP", "wrong server", "use hawksoft not ezyvet", "this should come from".

  • Skill create

    Srltas/claude-skills/skill-create Skill

    0★ repo

    Create a new skill in this claude-skills collection, from scaffolding through a lightweight trigger/dry-run test before commit. Use when the user wants to turn a workflow they just did into a reusable skill, capture a repeated process, or add a new skill to their toolbox. Applies this collection's conventions (domain prefixes, description-with-triggers, just install) plus skill-authoring practices from Anthropic's skill-creator (progressive disclosure, trigger-tested descriptions, bundled scripts/references/assets). Triggers on phrases like 'make this a skill', 'create a skill for this', 'save this as a skill', 'add a new skill', 'turn this workflow into a skill'.

  • Hook lab

    lumizone/social-content-os-claude-skill/skills/hook-lab Skill

    0★ repo

    Generate and rank scroll-stopping hooks for a single content idea. Use when the user wants hooks, openers, first lines, video first-3-seconds, thumbnail text, or says "wymysl hooki", "hook na X", "10 wariacji haka", "scroll-stopper", "give me hooks", "first line for this post", "how should I open this". Takes one idea and produces 10+ hook variations across the full taxonomy (Curiosity Gap, Pattern Interrupt, Challenge a Common Belief, Negation, Number/Result, Before/After), each tagged with its technique and why it works, then ranks them and recommends a top 3 with the visual/written/audio treatment for the winner.

  • Rag eval guardrails

    NeuralMedic-DE/claude-skills/rag-eval-guardrails Skill

    0★ repo

    Build a verified eval harness for a RAG/LLM feature plus PII/PHI-leakage guardrails, gated by checks that actually run. Scores a precomputed predictions file (so it runs with ZERO API access) on groundedness, citation validity, retrieval hit@k, answer F1/exact-match, refusal rate, and latency; compares to config thresholds and a baseline to catch regressions; and fails the build on PII/PHI leakage. Use when the user wants to evaluate or regression-test an AI/RAG feature, measure hallucination/groundedness, add an eval gate to CI, or scan prompts/answers/logs for leaked identifiers. Triggers: "RAG evaluation", "LLM eval", "eval harness", "hallucination", "groundedness", "PII/PHI leakage", "guardrails", "regression testing for AI features".

  • Eval set

    robdasi/skills/eval-set Skill

    0★ repo

    Decide whether an automation, agent, or skill actually works, without pretending you run a test suite. Turns the output's contract into a schema, a pre-ship checklist of must-be-true invariants, and a completeness judge with a numeric gate, plus one regression case per bug already found. Use this before you trust a build, before a model upgrade silently breaks it, or before each unattended run. Produces a checks file and stops.

  • Llm evaluation

    SkillMedev/ai-engineer-toolkit/skills/llm-evaluation Skill

    0★ repo

    Builds evaluation harnesses for LLM products - golden datasets, deterministic checks, calibrated LLM-as-judge rubrics, and CI regression gates that turn "seems good" into a tracked number. Use when someone asks "how do I know this prompt change didn't break anything", "set up evals for my RAG pipeline", "is LLM-as-judge reliable", "why did quality drop after the model swap", or is shipping an LLM feature with no quality measurement. Do NOT use for classical ML model reporting (precision/recall, ROC curves, confusion matrices on trained classifiers) - use model-evaluation-report instead; do NOT use for analyzing online A/B experiments - use ab-test-analyzer instead.

    The same file is in 1 other repository

  • Prompt engineer

    SkillMedev/ai-engineer-toolkit/skills/prompt-engineer Skill

    0★ repo

    Turns a vague request into a structured, reliable prompt - role, context, task, format, failure handling, and few-shot examples - that produces consistent output across real inputs. Use when someone asks "why does my prompt give inconsistent results", "write a prompt for this task", "the model keeps breaking my JSON", "how do I stop prompt injection from user input", or is building any LLM feature whose prompt was written ad hoc. Do NOT use for converting a working prompt into a reusable agent skill - use prompt-to-skill instead; do NOT use for measuring whether a prompt change improved quality - use llm-evaluation instead.

    The same file is in 1 other repository

  • Project md

    0jg/skills/project-md Skill

    no license0★ repo

    Generates a project-specific CLAUDE.md file at project initialisation time. Use this skill whenever starting a new project, forking a repo, or when the user asks to set up Claude Code for a project. Always invoke this skill as part of the research-project-init workflow — do not write CLAUDE.md from memory. The generated file encodes project type, physics context, toolchain rules, directory conventions, agent configuration, and skill triggers so that all context persists across sessions and survives context compaction.

  • Aeo llm visibility audit

    saurabhshuklagrowisto/saurabh-ai-systems/claude-skills/aeo-llm-visibility-audit Skill

    0★ repo

    Measures whether a brand shows up when buyers ask AI assistants (ChatGPT, Perplexity, Gemini) about its category, and turns the gaps into a content plan. Takes a brand name and a set of real buyer questions, checks which brands each engine names, and returns a share-of-voice score, the competitors winning the answers, and the exact questions where the brand is invisible. Use to audit AI/answer-engine discoverability (AEO/GEO), to prioritise content that AI assistants will cite, or to track share of voice in AI answers over time.

  • Figmahandoff

    saurabhshuklagrowisto/saurabh-ai-systems/claude-skills/figmahandoff Skill

    0★ repo

    Use this skill EVERY TIME a Figma URL is mentioned or pasted in any message. Triggers on any figma.com link, any mention of "figma", or any request to implement/build/code a design. Before writing any code or reading any design, this skill intercepts the request and asks two required questions: (1) which tech stack — React or Shopify Liquid, and (2) whether to show the Figma design system first or proceed directly. Do NOT skip these questions. Do NOT write any code before both are answered. This is mandatory for all Figma-related implementation tasks at Acme.

  • Req eval

    Frog1205/OPC-Skills/skills/req-eval Skill

    0★ repo

    需求评估器——按项目类型用不同"镜头"评估一个需求/项目能不能做、值不值得做、报价合不合理。覆盖四类:①外包需求类(接单/甲方付费,重点:范围边界+验收可测性+报价合理性+合同风险)②项目合作类(合伙/共建/分成,重点:权责分工+利益分配+退出机制)③自研商业产品(自己做来卖,重点:真伪需求+市场竞品+商业模式+MVP+ROI)④实用小工具类(自用/小范围提效,重点:有无现成替代+工时+最小实现)。触发条件:用户说"评估这个需求""这个需求能不能做""值不值得做""报价合不合理""项目可行性""接不接这个项目""需求清不清楚""范围边界在哪里",或明说"外包/接单/甲方付费""合作/合伙/共建/分成/股权""自研/做产品/创业/SaaS/卖钱""小工具/脚本/自动化/提效"并要求评估/分析/报价/可行性判断。产出默认 Markdown 快评(分段+表格),加 --word 联动 word-doc 技能出正式 Word,加 --brief 只要一页结论。严格遵循 CLAUDE.md 推理校验规则:结论前先列前提、重要结论自我反驳一次、标注置信度、不确定先问不瞎猜。