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

  • Ai seo

    cskwork/supermarketer-skill/vendor/marketingskills/skills/ai-seo Skill

    1★ repo

    When the user wants to optimize content for AI search engines, get cited by LLMs, or appear in AI-generated answers. Also use when the user mentions 'AI SEO,' 'AEO,' 'GEO,' 'LLMO,' 'answer engine optimization,' 'generative engine optimization,' 'LLM optimization,' 'AI Overviews,' 'optimize for ChatGPT,' 'optimize for Perplexity,' 'AI citations,' 'AI visibility,' 'zero-click search,' 'how do I show up in AI answers,' 'LLM mentions,' 'optimize for Claude/Gemini,' 'llms.txt,' 'OKF,' 'Open Knowledge Format,' 'knowledge bundle,' or 'agent-readable site.' Use this whenever someone wants their content to be cited or surfaced by AI assistants and AI search engines. For traditional technical and on-page SEO audits, see seo-audit. For structured data implementation, see schema.

    The same file is in 4 other repositories

  • Sql agent

    zhangtao272820/AI_Agent/DB_Agent/.trae/skills/sql_agent Skill

    no license1★ repo

    处理复杂/开放式业务查询与多表分析。当问题超出固定技能、需要多步推理或跨表关联时调用。

  • Register agent

    uptopia-team/legend-of-base-agent-skills/skills/register-agent Skill

    1★ repo

    Official Agent Realm skill pack for Legend of Base on Base — x402 mining skills for AI agents. Sponsored by Uptopia.

  • Tool review gate

    wde123sadw/agent-preflight/skills/tool-review-gate Skill

    1★ repo

    Review a proposed skill, plugin, MCP server, CLI, package, API, connector, application, or service before installation or use. Use when a new capability may execute code, access files or accounts, receive project data, require secrets, cost money, change configuration, or create external effects. Present a decision-ready approval card and wait for explicit, scoped user approval.

  • Using agent preflight

    wde123sadw/agent-preflight/skills/using-agent-preflight Skill

    1★ repo

    Route non-trivial, ambiguous, costly, long-running, or high-risk agent work through adaptive intent clarification, capability gap analysis, tool discovery and review, and context budgeting. Use at the start of build, change, diagnosis, research, creative, data, or automation tasks when wrong assumptions, missing capabilities, large context, irreversible actions, or meaningful time or money are plausible. Skip trivial self-contained questions and clear low-risk mechanical work.

  • 507 handoff

    ssdiwu/507-skills/common/507-handoff Skill

    1★ repo

    会话移交:中断、换会话、压缩上下文或转交给另一 Agent 时,在对话中生成脱敏的临时交接摘要,不写文件。Use when user says handoff, 交接, 交接一下, 移交, 换会话, 换个会话, 转交, 压缩上下文, 上下文太长了, 给下一个 agent 摘要.

  • Handoff

    eduardkumskyi/octo/skills/handoff Skill

    no license1★ repo

    Write .claude/handoff.md so any future session can resume: current goal, done/remaining tasks, key decisions and assumptions, gotchas discovered, next step. The context-restore hook injects its head after compaction.

  • Skill

    eduardkumskyi/octo/skills/skill Skill

    no license1★ repo

    Author a new skill, agent, or hook - into the octo plugin repo or a host project's .claude/. Knows the current frontmatter formats, hook events, and plugin layout; scaffolds, dry-runs, and commits.

  • Multi agent coordination

    andr-ca/agentharness/.claude/skills/multi-agent-coordination Skill

    1★ repo

    Use when two or more agent sessions may work on the same repository concurrently — covers the per-feature lock-file protocol, stale-lock detection, worktree isolation rules, and what to do when a feature is already locked.

  • Port agent config

    andr-ca/agentharness/.claude/skills/port-agent-config Skill

    1★ repo

    Use when asked to port, migrate, or add equivalent agent instructions, skills, or custom sub-agents for a different coding tool than the one already configured — e.g. "add Cursor support from our CLAUDE.md", "we're switching from Cursor to Codex, port our rules", "make this work for Copilot too", "port our review subagent to Codex". Covers both agentharness-linked projects (use the real generators, don't hand-write) and plain projects with hand-authored config (port by hand, same principles).

  • Subagent orchestration

    ArcanEdge-AI/codex-agent-playbook/skills/subagent-orchestration Skill

    1★ repo

    Use when a coding task may benefit from Codex subagents for planning, engineering, review, testing, or documentation lookup. Helps the main agent delegate bounded work and verify outputs before accepting them.

  • Harness prompt

    atlax-tech/harness-armor/skills/harness-prompt Skill

    no license1★ repo

    Generate source-grounded execute, test, and review engineering prompts for every step in an existing implementation plan, with independent roles and explicit file/verification boundaries. Use when a repository already has product facts, Harness constraints, and a real roadmap or plan. Do not use to improve generic chat prompts, invent a plan, implement code, or let executors self-approve.

  • Setup agentic os

    giacomoguidotto/skills/skills/setup-agentic-os Skill

    1★ repo

    🧩 Curated, auditable agent skills for knowledge, mastery, career, and cross-system orchestration.

    The same file is in 1 other repository

  • Ai search geo

    whatsuppiyush/god-of-skills/seo/ai-search-geo Skill

    1★ repo

    Get recommended by AI search (ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude) using GEO / AEO: audit per-model visibility, structure content so LLMs cite it, earn off-site mentions and entity consistency, then run it as a monthly cadence with GA4 attribution. Use whenever the user wants to appear in AI answers, asks about GEO, AEO, "generative engine optimization", LLM visibility, ChatGPT/Perplexity recommendations, AI Overviews, being cited by AI, "why does ChatGPT recommend my competitor", ghost citations, llms.txt, or tracking AI-referred traffic that hides as direct. Trigger phrases: "show up in ChatGPT", "AI search optimization", "get cited by Perplexity", "we rank on Google but AI never mentions us", "AI visibility audit", "which model recommends us".

  • Destructive git hook

    jeremyinthebay/relay-skills/skills/destructive-git-hook Skill

    1★ repo

    Install a PreToolUse hook that mechanically blocks git reset --hard, push --force, branch -D and other work-destroying commands — a gate an agent cannot talk its way past, unlike an instruction in CLAUDE.md. Use whenever an agent runs unattended or with --dangerously-skip-permissions, before letting an agent near a repo you care about, after an agent has proposed or run a destructive git command, or when someone asks how to stop an AI from deleting their work.

  • Llm feature adversarial audit

    jeremyinthebay/relay-skills/skills/llm-feature-adversarial-audit Skill

    1★ repo

    Adversarially audit an LLM-backed product feature for the exploit classes users actually find — client-controlled conversation history injecting false facts, XSS on rendered model output or user input, cost/kill-switch adequacy under variable payload size, auth/metering bypass, and secret handling. Use before shipping or reviewing an AI chat/assistant feature, before trusting client-supplied conversation history or documents, when adding a budget cap or kill switch to an LLM feature, or when someone asks "is our AI feature safe," "can this be jailbroken," or "can a user make it say/do something it shouldn't."

  • Two claude relay

    jeremyinthebay/relay-skills/skills/two-claude-relay Skill

    1★ repo

    Set up a two-agent autonomous loop where a planning agent writes briefs, an executing agent builds and opens PRs, and the planner independently verifies and merges — running unattended for hours. Use this whenever the user wants an agent to work autonomously, run overnight, keep working while they're away, self-review its own code, build a multi-agent or planner/executor setup, or asks how to let Claude Code run without supervision. Also use when someone asks "can I have an agent just keep going on its own" or wants to automate a build/review/merge cycle.

  • Rpa module asset agent

    shuhaolin63-hash/Big-Data-and-Accounting-RPA-skills/.trae/skills/rpa-module-asset-agent Skill

    1★ repo

    RPA模块资产智能体:双模式作业生成+资产库管理。当用户需要从零生成RPA作业(无源码场景)、需要补齐缺失素材(表格/图片/代码模板)、需要批量替换本机路径、需要调用资产库模板生成标准作业时触发。核心资产库直接引用f:\RPA(1)下的机器人文件夹作为真实资产。

  • Rpa teacher agent

    shuhaolin63-hash/Big-Data-and-Accounting-RPA-skills/.trae/skills/rpa-teacher-agent Skill

    1★ repo

    RPA作业老师教学智能体:框选标准化教学+作业合规指导+考点讲解。当用户需要从零完成RPA作业、不懂框选操作、需要框选可视化交底、需要ID合规教学、需要作业验收标准说明时触发。双场景:有源码时讲解代码逻辑,无源码时手把手指导框选操作。

  • Big Data and Accounting RPA skills

    shuhaolin63-hash/Big-Data-and-Accounting-RPA-skills Skill

    1★

    RPA学生多智能体作业系统总控Skill:统一调度三大子智能体(老师/破障/资产),覆盖有源码改作业、无源码做新作业双场景。当用户需要完成RPA作业、处理源码ID风险、框选操作教学、从零生成作业时触发。根据用户需求自动路由到对应的子智能体。

  • Build knowledge base

    dineshrevunuru/SuperSkills/build-knowledge-base Skill

    1★ repo

    Capture reusable learnings, decisions, and patterns into a findable personal knowledge base — right home, tagged, linked — so future-him retrieves instead of re-deriving. Use when Dinesh says 'note this down', 'save this for later', 'where do I keep this', 'I'll forget this', 'did I solve this before', 'knowledge base', 'second brain', 'my notes', 'capture this learning', or when a hard-won fix / decision / pattern would otherwise die in a chat log. Owns personal knowledge (capture → link → retrieve); research findings live in build-research-repository.

  • Define ux success metrics

    dineshrevunuru/SuperSkills/define-ux-success-metrics Skill

    1★ repo

    Defines how a design's success will be measured: HEART goals→signals→metrics, SUS/SEQ/NPS instrument selection, UX benchmarking waves, and analytics measurement plans with macro/micro conversions. Use when asked 'how do we measure success', 'what metrics should we track', 'define KPIs for this feature', 'set up HEART', 'should we use NPS', 'benchmark our UX', 'write a measurement plan', or before any launch that will later be judged. NOT for analyzing usability findings (analyze-usability-data), designing A/B tests (use-quantitative-evidence), or measuring AI output quality (write-ai-evals).

  • Design agentic ux

    dineshrevunuru/SuperSkills/design-agentic-ux Skill

    1★ repo

    Designs the surfaces of an AI that ACTS — multi-step plan display, run progress, streaming and think-time states, interrupt/cancel/undo, human-in-the-loop approval gates and review queues, memory visibility and control, and non-determinism affordances (regenerate, versions). Use when an AI works over multiple steps or on the user's behalf: 'show the agent's plan', 'design the approval flow', 'what does the user see while it's working', 'add a regenerate button', 'let users see or edit what it remembers', 'the agent did something without asking', 'progress UI for a long-running AI task', 'design the review queue'. Owns the acting-AI experience; what happens when the AI is WRONG (hallucination, refusal, error states) belongs to design-ai-trust-and-failure-states.

  • Design ai trust and failure states

    dineshrevunuru/SuperSkills/design-ai-trust-and-failure-states Skill

    1★ repo

    Design the user-facing reliability layer of any AI product: uncertainty and confidence display, provenance/citations, hallucination and sycophancy mitigations, explainability calibrated to stakes, and a named cause + recovery path for EVERY failure — refusal, timeout, content filter, context overflow, mic/STT death, TTS death, dead live-embed. Use when asked 'what happens when the AI is wrong/fails/refuses', 'design the error/fallback/empty state for the AI', 'how do we show confidence or sources', 'the model hallucinated', 'the mic/voice broke', or before shipping ANY AI feature — no AI surface ships with only a happy path. Owns failure UX; measuring failure belongs to write-ai-evals.

  • Engineer production prompts

    dineshrevunuru/SuperSkills/engineer-production-prompts Skill

    1★ repo

    Writes, structures, versions, and regression-guards system prompts as production artifacts — architecture, few-shot examples, register constraints, model pinning, and change discipline. Use when writing or reviewing a system prompt for a shipped AI feature; when someone says 'tweak the prompt', 'the model keeps saying X', 'add an example to the prompt', 'change the wording', 'switch the model', 'why did the behavior change', or 'is this prompt good'; when encoding voice/safety/tie-break rules into prompt text; or when a prompt change is about to ship without a regression run. Distilled from Dinesh's shipped prompts (the Boo router, the Saathi shopkeeper turn).

  • Git guardrails

    dineshrevunuru/SuperSkills/git-guardrails Skill

    1★ repo

    Install a Claude Code PreToolUse hook that BLOCKS irreversible git commands before an agent runs them — hard reset, clean -f, force-push, branch -D, checkout/restore ., history rewrites. The software enforcement of ARIA's Tier-3 'never auto': no destructive, un-undoable git op self-authorized by the agent. Use when setting up a new repo or machine, when 'add a git guard / git safety hook', 'block dangerous git', 'stop the agent from force-pushing / hard-resetting', 'prevent destructive git', or after an agent nukes uncommitted work. Infra: install once, on by default everywhere.

  • Handoff context

    dineshrevunuru/SuperSkills/handoff-context Skill

    1★ repo

    Compact a conversation into a handoff another agent can resume from — decisions + state + open threads + the ONE next action, never a transcript. Shaped to the ARIA Hub session-report schema so a handoff FEEDS the registry instead of inventing a parallel format. Use when a session is ending, context is running low, work passes to a fresh agent or sub-agent, or asked 'hand this off', 'write a handoff', 'summarize this for the next session', 'compact this so someone can pick it up', 'session summary for the Hub', 'wrap up so I can continue tomorrow'. Mode-switch: a 3-line quick pass vs a full state dump.

  • Write ai evals

    dineshrevunuru/SuperSkills/write-ai-evals Skill

    1★ repo

    Designs and runs evals for any AI feature the way Dinesh does — golden sets, grading rubrics, LLM-as-judge, hard safety gates, failure-mode taxonomies fed back into structural fixes, and AI-quality product metrics (acceptance, regeneration, edit-distance). ALSO owns the pre-design model capability assessment: what can this model actually do for this use case, cost-latency-quality tradeoffs, model-tier and provider choice. Use when asked 'which model should we use', 'is the small/cheap model good enough', 'how do we test the AI', 'write test cases for the bot', 'is it hallucinating', 'did the prompt change break anything', 'how do we know the AI is good', or BEFORE designing any LLM, voice, agent, or recommender feature. Measures failure; the user-facing failure UX belongs to design-ai-trust-and-failure-states.

  • Mcp builder

    nguyenthdat/opencode-starter/skills/mcp-builder Skill

    1★ 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 in any official SDK language: Python (FastMCP), TypeScript/Node, Go, Rust (rmcp), Java, Kotlin, C#/.NET, PHP, Ruby, or Swift.

  • Skill creator

    nguyenthdat/opencode-starter/skills/skill-creator Skill

    1★ repo

    Create, edit, evaluate, package, and improve opencode skills. Use when the user asks to create a new skill, convert a workflow into a skill, refine a SKILL.md, improve a skill description, run skill evals, benchmark with-skill vs baseline behavior, package a .skill artifact, or when the harness skill needs a dedicated skill-building workflow.

  • Openclaw skill

    doteyeso-ops/vibes-coded-agent-connector/src/openclaw-skill Skill

    1★ repo

    OpenClaw + npm vibes-coded-agent-connector: register agents, listings, Solana escrow jobs, prepaid /start + X-Vibes-Key, Operator Interrupt (X-Operator-Notify), hosted uploads, checkout, receipts, affiliates. Reclaim SOL; Hermes well-known on connector site.

  • Consolidate memory

    furkangonel/cowrangler/bundled_skills/consolidate-memory Skill

    1★ repo

    Reflective pass over your memory files — merge duplicates, fix stale facts, prune the index. Trigger for: - [TODO: Add specific triggers for when to use this skill] Don't trigger for: - [TODO: Add anti-triggers for when NOT to use this skill]

  • Mcp builder

    furkangonel/cowrangler/bundled_skills/mcp-builder Skill

    1★ 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).

  • Vector databases

    furkangonel/cowrangler/bundled_skills/mlops/vector-databases Skill

    1★ repo

    Vector database operations — embed, store, search, and build RAG pipelines.

  • Prompt engineer

    furkangonel/cowrangler/bundled_skills/prompt-engineer Skill

    1★ repo

    Turn vague instructions into precision-crafted prompts and system messages that get the best out of any LLM

  • Skill creator

    furkangonel/cowrangler/bundled_skills/skill-creator Skill

    1★ repo

    Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.

  • Ansi safe border caption embedding

    vishuwa2004/cskill-agents/agents/claude-code/skills/ansi-safe-border-caption-embedding Skill

    1★ repo

    Embed captions into terminal borders while preserving alignment, clipping, and color/dim semantics.

    The same file is in 1 other repository

  • Auto memory tool permission fencing

    vishuwa2004/cskill-agents/agents/claude-code/skills/auto-memory-tool-permission-fencing Skill

    1★ repo

    Gate every tool call in the auto-memory extract agent so only safe read-only actions and targeted writes execute.

    The same file is in 1 other repository

  • Built in agent feature gating

    vishuwa2004/cskill-agents/agents/claude-code/skills/built-in-agent-feature-gating Skill

    1★ repo

    Gate built-in agents based on feature flags and entrypoint types so SDK sessions stay lean.

    The same file is in 1 other repository

  • Bundled and mcp turn zero skill filter

    vishuwa2004/cskill-agents/agents/claude-code/skills/bundled-and-mcp-turn-zero-skill-filter Skill

    1★ repo

    For first-turn skill listings, prefer a small bundled+MCP subset and fall back to bundled-only when the filtered set is still too large.

    The same file is in 1 other repository

  • Cache safe stop hook snapshotting

    vishuwa2004/cskill-agents/agents/claude-code/skills/cache-safe-stop-hook-snapshotting Skill

    1★ repo

    Persist a cache-safe stop-hook context snapshot only for main session queries so later helpers can resume from stable state without fork pollution.

    The same file is in 1 other repository

  • Canonical worktree path normalization

    vishuwa2004/cskill-agents/agents/claude-code/skills/canonical-worktree-path-normalization Skill

    1★ repo

    Resolve repo-relative paths against the canonical checkout root so shared state stays stable across Git worktrees.

    The same file is in 1 other repository

  • Conversation engine persistent session state

    vishuwa2004/cskill-agents/agents/claude-code/skills/conversation-engine-persistent-session-state Skill

    1★ repo

    Keep session-level signals (messages, read caches, memory loads) alive across turns while letting per-turn caches reset for correctness.

    The same file is in 1 other repository

  • Fast worktree reuse with head pointer checks

    vishuwa2004/cskill-agents/agents/claude-code/skills/fast-worktree-reuse-with-head-pointer-checks Skill

    1★ repo

    Resume existing worktrees quickly by checking stored HEAD pointers before fetching or recreating.

    The same file is in 1 other repository

  • Forked auto memory extraction

    vishuwa2004/cskill-agents/agents/claude-code/skills/forked-auto-memory-extraction Skill

    1★ repo

    Run a forked agent that captures stable session context and writes auto memories when the main agent finishes a turn.

    The same file is in 1 other repository

  • Hook context attachment injection

    vishuwa2004/cskill-agents/agents/claude-code/skills/hook-context-attachment-injection Skill

    1★ repo

    Surface hook-provided startup context by packaging it as metadata attachments before the agent's first request.

    The same file is in 1 other repository

  • Local and mcp skill name deduplication

    vishuwa2004/cskill-agents/agents/claude-code/skills/local-and-mcp-skill-name-deduplication Skill

    1★ repo

    Merge local and MCP skill sources into one listing surface and dedupe by stable command name before announcing them.

    The same file is in 1 other repository

  • Main agent vs background memory extraction exclusion

    vishuwa2004/cskill-agents/agents/claude-code/skills/main-agent-vs-background-memory-extraction-exclusion Skill

    1★ repo

    Skip the background memory extractor when the main agent already persisted memory during the same turn to avoid duplicates and race hazards.

    The same file is in 1 other repository

  • Main thread empty string agent key normalization

    vishuwa2004/cskill-agents/agents/claude-code/skills/main-thread-empty-string-agent-key-normalization Skill

    1★ repo

    Normalize the main thread to a concrete empty-string key so agent-scoped state maps treat main and forked threads uniformly.

    The same file is in 1 other repository

  • Mark after filter memory deduplication

    vishuwa2004/cskill-agents/agents/claude-code/skills/mark-after-filter-memory-deduplication Skill

    1★ repo

    Filter prefetched memory attachments against cumulative read state before mutating that state, so async surfacing does not dedupe itself away.

    The same file is in 1 other repository

  • Mcp skill vs prompt discovery filtering

    vishuwa2004/cskill-agents/agents/claude-code/skills/mcp-skill-vs-prompt-discovery-filtering Skill

    1★ repo

    Ensure SkillTool only exposes MCP skills when they are registered as prompt-based commands and dedupe them against local commands.

    The same file is in 1 other repository

  • Media block compaction prep

    vishuwa2004/cskill-agents/agents/claude-code/skills/media-block-compaction-prep Skill

    1★ repo

    Sanitize user messages by replacing images, documents, and other heavy media blocks with lightweight placeholders before compaction.

    The same file is in 1 other repository

  • Memory directory exists guidance

    vishuwa2004/cskill-agents/agents/claude-code/skills/memory-directory-exists-guidance Skill

    1★ repo

    Ensure memory directories exist once and surface consistent guidance so agents skip redundant mkdir calls.

    The same file is in 1 other repository

  • Memory entrypoint with index discipline

    vishuwa2004/cskill-agents/agents/claude-code/skills/memory-entrypoint-with-index-discipline Skill

    1★ repo

    Keep MEMORY.md terse by capping entries, warning when caps hit, and pointing each memory to its own file.

    The same file is in 1 other repository

  • Nested memory trigger consume and clear

    vishuwa2004/cskill-agents/agents/claude-code/skills/nested-memory-trigger-consume-and-clear Skill

    1★ repo

    Process every nested-memory trigger before wiping the trigger set so subsequent iterations only see new paths.

    The same file is in 1 other repository

  • Per agent skill listing dedupe

    vishuwa2004/cskill-agents/agents/claude-code/skills/per-agent-skill-listing-dedupe Skill

    1★ repo

    Keep skill-listing deltas scoped per agent or thread so each agent gets its own first-turn announcement surface.

    The same file is in 1 other repository

  • Plugin only agent surface gating

    vishuwa2004/cskill-agents/agents/claude-code/skills/plugin-only-agent-surface-gating Skill

    1★ repo

    Skip agent-declared MCP servers when strict plugin-only mode is active unless the agent is admin-trusted, keeping the session secure.

    The same file is in 1 other repository

  • Prevent continuation stop hook contract

    vishuwa2004/cskill-agents/agents/claude-code/skills/prevent-continuation-stop-hook-contract Skill

    1★ repo

    Return a narrow stop-hook result that can halt the query loop without throwing, while still surfacing blocking errors separately.

    The same file is in 1 other repository

  • Prevent continuation stop hook messaging

    vishuwa2004/cskill-agents/agents/claude-code/skills/prevent-continuation-stop-hook-messaging Skill

    1★ repo

    Collect stop-hook progress/errors, emit a structured prevented-continuation attachment, and summarize the block for users and logs.

    The same file is in 1 other repository

  • Progressive compaction prompt

    vishuwa2004/cskill-agents/agents/claude-code/skills/progressive-compaction-prompt Skill

    1★ repo

    Drive compaction summaries through structured prompts that sequence naming, artifacts, and invocation guidance before stripping analysis tags.

    The same file is in 1 other repository