agentleFS

Claude skills and agent skills

Skills real projects publish on GitHub, most starred first. Each one says what it will make an agent do before you copy it.

Worked for most · soon
MicrosoftReads credentialsSkill

azure-openai-to-responses

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/azure-openai-to-responses/SKILL.md

權威指南 — 請嚴格遵循 本技能將使用 Azure OpenAI Chat Completions 的 Python 代碼庫 遷移到統一的 Responses API。請精確遵循這些指令。 不得自行即興映射參數或創建 API 形態。 當用戶想要: - 將 Python 應用從 Azure OpenAI Chat Completions 遷移到 Responses API - 升級 Python OpenAI SDK 使用方式,改用針對 Azure OpenAI 的最新 API 形態 - 為需要 Azure Responses 的 GPT-5 或更新模型準備 Python 代碼 - 從 AzureOpenAI / AsyncAzureOpenAI 切換到使用標準的 OpenAI / AsyncOpenAI 客戶端,搭配 v1 端點 - 修正與 AzureOpenAI 建構子或 api_version 相關的棄用警告 遷移前,請確認您的 Azure OpenAI 部署支援 Responses API。 <strong>注意</strong>:Azure OpenAI 上 max_output_tokens 的<strong>最小值為 16</strong>。低於 16…

Steady33 mentions in 2 weeks

1 project uses a version of this

79k4mo agoReviewMountCreate a free account to mount
MicrosoftSkill

deploying-scalable-agents

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/deploying-scalable-agents/SKILL.md

第16課 - 部署可擴充代理 的配套技能。 用它來協助學習者將代理從原型轉移到可擴展、可觀察的生產部署。 將每一項建議都根植於課程內容和可運行筆記本;不得自行創造 Foundry API。 當學習者想要: - 將代理部署到 Microsoft Foundry 作為<strong>託管代理</strong>,並使其版本化/可觀察。 - 在<strong>用戶端託管、託管代理及代理工作流程</strong>部署模式中進行選擇。 - 添加<strong>模型路由</strong>、<strong>回應快取</strong>或<strong>有界並行性</strong>來控制延遲與成本。 - 添加一個<strong>評估門檻</strong>,以防止錯誤的代理版本被部署。 - 添加一個<strong>人機介入的核准</strong>步驟,針對高風險操作。 - 以<strong>OpenTelemetry</strong>追蹤為代理進行生產環境的觀察性儀表。 - 進行<strong>冒煙測試</strong>作為快速的部署後門檻。 生產代理主要是在模型之外的運作框架(約佔80%), 而非模型本身。將每個建議對應到以下關注點之一: create → version → evaluate (gate) → deploy hosted → observe online → collect failures → repeat。 離線評估是一道門檻,而非事後考量 — 版本不會上線,除非通過此門檻。 在線可觀察性會將真實失敗反饋至離線測試集。 從筆記本指向學習者這些 16-python-agent-framework.ipynb: 部署後,驗證端點確實有回應(綠燈部署仍可能無聲無息)。 透過 AI Smoke Test 動作使用 .github/workflows/smoke-test.yml 搭配 tests/ 中的目錄。執行器會將每個 提示以 POST 發送至 POST {project_endpoint}/agents/{agent_name}/endpoint/protocols/openai/responses, 並對回覆文字斷言。身份需具備 Foundry 專案範圍內的<strong>Azure AI User</strong>角色;Token 受眾必須是 https://ai.azure.com/。

Steady33 mentions in 2 weeks

No other project uses it yet

79k4mo agoReviewMountCreate a free account to mount
MicrosoftSkill

jupyter-notebook

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/jupyter-notebook/SKILL.md

當使用者要求建立、搭建或編輯 Jupyter 筆記本(`.ipynb`)用於實驗、探索或教學時使用;優先使用隨附範本並執行輔助腳本 `new_notebook.py`

Steady33 mentions in 2 weeks

17 projects use a version of this

Often used withfind-skills

79k4mo agoReviewMountCreate a free account to mount
MicrosoftSkill

local-ai-agents

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/local-ai-agents/SKILL.md

伴隨 Lesson 17 – Creating Local AI Agents 的輔助技能。 用於協助學習者打造一個能獨立在自己機器上推理、調用工具及搜尋文件的代理 — 完全無雲端推理。所有建議都基於課程內容與可執行筆記本。 當學習者希望: - 完全在設備端執行代理,為了隱私、成本或離線需求。 - 使用 Foundry Local 本地部署模型,並透過 OpenAI 兼容端點連接。 - 使用 Qwen 函數呼叫 模型來驅動可靠的本地工具調用。 - 新增 本地 RAG(Chroma)或 本地 MCP 伺服器。 - 設計一個 <strong>混合</strong> 本地/雲端路由策略。 SLM 以隱私、成本與離線操作換取廣度。致勝策略是:讓 SLM 負責協調,讓工具執行繁重任務。 模型不需要<em>了解</em>整個程式碼庫,只需知道何時呼叫 read_file 和 search_docs。這符合 SLM 的強項(例如界定決策如工具選擇),避開其弱項(廣泛知識、長距多跳推理)。 8GB 記憶體為實際最低需求;GPU/NPU 有助於性能但非必需。 指引學習者閱讀筆記本 17-local-agent-foundry-local.ipynb: 這呼應了 Lesson 16 的模型路由概念,將工作站列為其中一條路徑。優先設計可回退至本地的方案,讓代理效能降低而非直接失效。

Steady33 mentions in 2 weeks

No other project uses it yet

79k4mo agoReviewMountCreate a free account to mount
MicrosoftSkill

microsoft-docs

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/microsoft-docs/SKILL.md

Microsoft 技術生態系統的研究技能。涵蓋 learn.microsoft.com 以及存在於其之外的文檔(VS Code、GitHub、Aspire、Agent Framework 倉庫)。 使用這些工具搜尋 learn.microsoft.com 上的所有內容 — 包括 Azure、.NET、M365、Power Platform、Agent Framework、Semantic Kernel、Windows 等。這是大多數 Microsoft 文件查詢的主要工具。 在需要完整教學、所有配置選項或搜尋摘要被截斷時,請於搜尋後使用 microsoft_docs_fetch。 以下類別存在於 learn.microsoft.com 之外。請改用指定工具。 Aspire 文件位於 aspire.dev,非 Learn。最佳工具取決於你的 Aspire CLI 版本: CLI 13.2+(推薦)— Aspire MCP 伺服器包含內建的文件搜尋工具: 這些功能隨 Aspire CLI 13.2 版本發布(PR #14028)。更新指令:aspire update --self --channel daily。 CLI 13.1 — MCP 伺服器提供整合查詢(list_integrations、get_integration_docs),但<strong>不提供</strong>文件搜尋。請使用 Context7 作為備用: VS Code 文件位於 code.visualstudio.com,非 Learn。 GitHub 文件位於 docs.github.com 與 cli.github.com。 Agent Framework 教學在 learn.microsoft.com(使用 microsoft_docs_search),但 GitHub 倉庫 提供的 API 細節往往比已發布文件更新 — 特別是 DevUI REST API 參考、CLI 選項及…

Steady33 mentions in 2 weeks

12 projects use a version of this

79k4mo agoReviewMountCreate a free account to mount
MicrosoftReads credentialsSkill

testing-course-samples

microsoft/ai-agents-for-beginners/translations/zh-TW/.agents/skills/testing-course-samples/SKILL.md

驗證課程筆記本和程式範例是否能在即時的 Microsoft Foundry / Azure OpenAI 環境中執行。此儲存庫提供一個運行器,位於 scripts/validate-notebooks.ps1, 它會無頭方式執行每個 Python 筆記本並印出通過/失敗矩陣。 此腳本會寫出執行後的複本、每筆記本的日誌,及 results.json 於 $env:TEMP\aiab-nbval,並以失敗數量作為退出碼。 短暫失敗(共用訂閱 HTTP 429 限流、偶發的 AzureCliCredential 令牌異常,或逾時)會自動重新嘗試 (-Retries,預設 2 次,搭配 -RetryDelaySeconds 回溯,預設 20 秒)。若 模型部署經常出現 429,請檢查訂閱的 GlobalStandard TPM 配額(az cognitiveservices usage list -l <region>)— 單一部署容量增加並無助於訂閱<em>整體</em>配額耗盡的情況。

Steady33 mentions in 2 weeks

No other project uses it yet

79k4mo agoReviewMountCreate a free account to mount
MicrosoftSkill

update-deps

microsoft/graphrag/.agents/skills/update-deps/SKILL.md

Update this GraphRAG uv-workspace monorepo's dependencies and repair the resulting breakages so every check and test passes again. Use when asked to update, upgrade, or bump dependencies/packages/versions, refresh or regenerate the lockfile, migrate to a new major of pandas/numpy/pydantic/pyarrow, resolve dependency-related test or lint failures, or run a "dependency sweep" — even if the user only names a single package. Covers editing pyproject.toml version specifiers across the root dev-deps and every packages/* member, re-locking with uv, and fixing code and tests for library API changes (e.g. pandas 3.0). USE FOR: update dependencies, upgrade packages, bump versions, dependency sweep, uv lock, uv sync, refresh lockfile, migrate pandas/numpy, "deps broke the tests", pyproject bump.

Steady3 mentions in 2 weeks

20 projects use a version of this

37k3mo agoReviewMountCreate a free account to mount
MicrosoftSkill

analyze-ci-failures

microsoft/vcpkg/.github/skills/analyze-ci-failures/SKILL.md

Analyze collected vcpkg CI failure evidence, classify regressions and known failures, identify root causes, recommend actions, and generate a report by package and triplet.

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

code-review

microsoft/vcpkg/.github/skills/code-review/SKILL.md

Review microsoft/vcpkg pull requests in GitHub Copilot code review. Check port correctness, packaging, dependencies, licenses, and consumer integration; validate examples when the review time budget permits.

Often used withclose-specadd-new-modelcompare-screenshots

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

evaluate-new-port

microsoft/vcpkg/.github/skills/evaluate-new-port/SKILL.md

Audit a vcpkg port locally. Reads the port metadata and build recipe, installs the port, inspects extracted sources and installed files, and produces a report covering license risks, vendored code, optional dependencies, and other port review suggestions.

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

fetch-vcpkg-ci-logs

microsoft/vcpkg/.github/skills/fetch-vcpkg-ci-logs/SKILL.md

Fetch vcpkg Azure DevOps metadata, failed-step logs, and selected artifacts. Use for acquiring CI evidence, not diagnosing or reporting regressions; use analyze-ci-failures for those tasks. Skip when supplied logs are sufficient.

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

fix-reviewed-vcpkg-pr

microsoft/vcpkg/.github/skills/fix-reviewed-vcpkg-pr/SKILL.md

Address findings in an existing report produced by review-vcpkg-pr or review-vcpkg-prs-today, then push fixes for the user to submit to the contributor. Use only for follow-up fixes to those reviews, including batches. Do not use for generic bug fixes, issue resolution, or PRs without an existing review report. Never open a PR.

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

review-vcpkg-pr

microsoft/vcpkg/.github/skills/review-vcpkg-pr/SKILL.md

Review a microsoft/vcpkg pull request end-to-end.

28k20d agoReviewMountCreate a free account to mount
MicrosoftReads credentialsSkill

review-vcpkg-prs-today

microsoft/vcpkg/.github/skills/review-vcpkg-prs-today/SKILL.md

Review open non-draft microsoft/vcpkg pull requests updated in the last 30 days. Use when asked for batch triage, per-PR review reports, an index grouped by verdict, or grouping competing PRs that modify the same port.

28k20d agoReviewMountCreate a free account to mount
MicrosoftSkill

code-review

microsoft/onnxruntime/.github/skills/code-review/SKILL.md

Review ONNX Runtime pull requests, branches, commits, patches, and working-tree changes for actionable findings and missing tests.

Often used withclose-specadd-new-modelcompare-screenshots

22k3mo agoReviewMountCreate a free account to mount
MicrosoftSkill

contrib-op-shape-inference-memory-safety

microsoft/onnxruntime/.github/skills/contrib-op-shape-inference-memory-safety/SKILL.md

Audit and fix out-of-range output writes in ONNX Runtime operator shape-inference functions. Use when reviewing or fixing a contrib (or standard) op TypeAndShapeInference where a getNumOutputs() guard precedes a write to a higher output index - optional trailing outputs make a smaller output count schema-valid, so getOutputType(index) can run one past the declared outputs at Graph::Resolve.

22k3mo agoReviewMountCreate a free account to mount
MicrosoftSkill

cuda-attention-kernel-patterns

microsoft/onnxruntime/.github/skills/cuda-attention-kernel-patterns/SKILL.md

Patterns and pitfalls for the ONNX domain Attention operator (opset 23/24) CUDA implementation. Use when modifying the dispatch cascade in core/providers/cuda/llm/attention.cc, writing mask/bias CUDA kernels, debugging attention test routing, or adding features to the ONNX Attention op. NOT for contrib domain MultiHeadAttention/GroupQueryAttention.

22k3mo agoReviewMountCreate a free account to mount
MicrosoftSkill

cuda-cutlass-fmha-incremental-rebuild

microsoft/onnxruntime/.github/skills/cuda-cutlass-fmha-incremental-rebuild/SKILL.md

Use when rebuilding ONNX Runtime CUDA after editing CUTLASS fused-MHA headers (onnxruntime/contrib_ops/cuda/bert/cutlass_fmha/*.h such as kernel_forward.h or fmha_launch_template.h), or when a header edit "passed" an incremental build but test behavior did not change. Explains the nvcc depfile gotcha that produces stale Memory-Efficient-Attention (MEA) kernels and binaries, and how to force a correct recompile. Also covers disk-space frugality on shared GPU dev boxes.

22k3mo agoReviewMountCreate a free account to mount
MicrosoftPipes a download into a shellSkill

onnx-opset-bump-checklist

microsoft/onnxruntime/.github/skills/onnx-opset-bump-checklist/SKILL.md

Step-by-step checklist for bumping the pinned ONNX dependency / opset in ONNX Runtime (e.g. ONNX 1.21 / opset 26 → 1.22 / opset 27). Use when integrating a new ONNX release or release-candidate, updating the cmake/deps.txt onnx pin or the cmake/external/onnx submodule, regenerating cmake/patches/onnx/onnx.patch, raising kMaxSupportedOpset, or adding a new opset's CPU kernels. Covers the file taxonomy, archive-hash procedures, patch rebase/mirror rules, the RC→formal strategy, and the optimizer/EP gotchas that the automated audit script misses.

22k3mo agoReviewMountCreate a free account to mount
MicrosoftSkill

ort-build

microsoft/onnxruntime/.github/skills/ort-build/SKILL.md

Build ONNX Runtime from source. Use this skill when asked to build, compile, or generate CMake files for ONNX Runtime.

22k3mo agoReviewMountCreate a free account to mount
CLAUDE.md vs AGENTS.md

About skills

What is a Claude skill?

A folder with a SKILL.md file: a name, a description of when to use it, and instructions. Claude loads a skill only when the task matches its description.

How do I use one I find here?

Copy the folder into your project's .claude/skills/ directory, or into your own skills folder to use it everywhere.

What do the warnings mean?

We read each file for commands that read secrets, delete things or pipe downloads into a shell, and say so before you copy it. No warning is not a promise that a file is safe.

Which skills worked for people?

Open a skill to see its discussion. Reports from people and their agents are coming.