Data Extractor

10 skills inclus

• 01

  Postgres Best Practices

  Fait appliquer à l'agent les règles Postgres : schéma, index, RLS, pooling, verrous et pagination, avec preuve par EXPLAIN.+

  sf-postgres-best-practices/SKILL.md, de supabase/agent-skills, licence MIT.

  1---

  2name: sf-postgres-best-practices

  3description: "Applies Postgres rules for schema design, indexes, Row Level Security, connection pooling, lock…

  4---

  5

  6# Postgres Best Practices

  7

  8Write Postgres schema, SQL and policies that stay correct, secure and fast as data grows. The rules live in `…

  9

  10## When to use

  11

  12- Any DDL: new table, new column, type change, constraint, index, trigger, function, policy.

  13- Any SQL that will run in production against a table that can grow.

  14- Reviewing a migration or a pull request that touches the database.

  15- Diagnosing slowness, lock waits, connection errors, or data leaking across tenants.

  16
• 02

  SQL Tuning

  Fait diagnostiquer une requête lente par l'agent : lecture du plan EXPLAIN, une hypothèse, un index ou une réécriture, puis nouvelle mesure.+

  sf-sql-tuning/SKILL.md, de wshobson/agents, licence MIT.

  1---

  2name: sf-sql-tuning

  3description: Diagnoses and fixes slow SQL with a measured loop of execution plans, indexing, query rewrites a…

  4---

  5

  6# SQL Tuning

  7

  8Make a slow query fast with evidence. Never guess an index: read the plan, form one hypothesis, change one th…

  9

  10## When to use

  11

  12- A known slow query, endpoint or report.

  13- A plan (`EXPLAIN` output) to interpret.

  14- Choosing indexes for a new access pattern.

  15- Finding the worst queries on a database (`pg_stat_statements`, slow query log).

  16
• 03

  SQL Anti-pattern Review

  Fait relire du SQL par l'agent sur tout moteur : prédicats non indexables, OFFSET, NOT IN, sous-requêtes corrélées, avec réécriture et index.+

  sf-sql-antipatterns/SKILL.md, de github/awesome-copilot, licence MIT.

  1---

  2name: sf-sql-antipatterns

  3description: Reviews SQL code for performance anti-patterns that hold on any engine (MySQL, PostgreSQL, SQL S…

  4---

  5

  6# SQL Anti-pattern Review

  7

  8Scan SQL for patterns that waste work on every engine, explain the cost of each, and propose a rewrite plus t…

  9

  10## When to use

  11

  12- Reviewing a selection of SQL, a migration, a stored procedure, a repository's queries, or an ORM query log.

  13- The engine is MySQL, SQL Server, Oracle or SQLite, or you do not know it yet.

  14

  15Not for: interpreting a specific execution plan step by step (use `sf-sql-tuning`), or Postgres schema and RL…

  16
• 04

  Zero-Downtime Database Migrations

  Fait planifier à l'agent des migrations sans interruption : expand and contract, DDL sans verrou long, backfill par lots et retour arrière testé.+

  sf-db-migrations/SKILL.md, de wshobson/agents, licence MIT.

  1---

  2name: sf-db-migrations

  3description: Plans and writes database schema and data migrations that deploy without downtime or data loss,…

  4---

  5

  6# Database Migrations

  7

  8Change a live schema while old and new application code keep working. Every migration must be safe to run whi…

  9

  10## When to use

  11

  12- Any migration that will run against a database with real traffic or data you cannot lose.

  13- Reviewing a generated migration (`prisma migrate dev`, `makemigrations`, `drizzle-kit generate`) before it…

  14- Data transformations: splitting a column, merging tables, changing units or encodings.

  15- Moving data between databases or engines.

  16
• 05

  API and Interface Design

  Fait concevoir à l'agent des API stables : contrat d'abord, erreurs uniformes, validation aux frontières, pagination et idempotence correcte.+

  sf-api-design/SKILL.md, de addyosmani/agent-skills, licence MIT.

  1---

  2name: sf-api-design

  3description: Designs stable, hard-to-misuse interfaces (REST and GraphQL endpoints, module boundaries, TypeSc…

  4---

  5

  6# API and Interface Design

  7

  8Define the contract before the implementation, make the right call easy and the wrong call hard, and evolve i…

  9

  10## When to use

  11

  12- New endpoints, GraphQL types, SDK methods, module exports, component props.

  13- Changing an existing public interface (renaming, retyping, removing).

  14- Any state-changing operation that clients or queues may retry.

  15

  16For the OpenAPI document itself, use `sf-openapi-spec`. For framework wiring (Express, Fastify, NestJS), use…
• 06

  OpenAPI Specifications

  Fait écrire, générer et valider à l'agent une spec OpenAPI 3.1 fidèle au code, puis l'utiliser pour la doc, les SDK et les tests de contrat.+

  sf-openapi-spec/SKILL.md, de wshobson/agents, licence MIT.

  1---

  2name: sf-openapi-spec

  3description: Writes, generates, lints and maintains OpenAPI 3.1 specifications, design-first or from code (Fa…

  4---

  5

  6# OpenAPI Specifications

  7

  8Produce an OpenAPI 3.1 document that is valid, linted, accurate to the implementation, and useful to the peop…

  9

  10## When to use

  11

  12- Writing a spec before building an API (design-first).

  13- Extracting a spec from an existing codebase (code-first).

  14- Fixing validation or lint errors in a spec.

  15- Generating SDKs, mock servers or documentation from a spec.

  16- Migrating Swagger 2.0 or OpenAPI 3.0 to 3.1.
• 07

  Node.js Backend

  Fait construire à l'agent un backend Node.js prêt pour la prod : couches, validation Zod, erreurs centralisées, JWT, rate limit, logs et arrêt propre.+

  sf-node-backend/SKILL.md, de wshobson/agents, licence MIT.

  1---

  2name: sf-node-backend

  3description: Builds production-ready Node.js backend services with Express or Fastify in TypeScript, with lay…

  4---

  5

  6# Node.js Backend

  7

  8Build Node.js services that validate every input, fail with clean errors, keep secrets out of code, and survi…

  9

  10## When to use

  11

  12- Scaffolding a new Express or Fastify service, or adding routes, middleware, auth, caching or database acces…

  13- Reviewing a Node backend before production.

  14

  15Not for: designing the API contract itself (use `sf-api-design`), writing the OpenAPI document (use `sf-opena…

  16
• 08

  Stripe Payments

  Fait intégrer Stripe par l'agent sur les bonnes API : Checkout Sessions, webhooks pour l'exécution, taxes, Connect, clés restreintes et sandboxes.+

  sf-stripe-payments/SKILL.md, de stripe/ai, licence MIT.

  1---

  2name: sf-stripe-payments

  3description: "Guides Stripe integrations end to end: test environment choice (sandboxes), API selection (Chec…

  4---

  5

  6# Stripe Payments

  7

  8Build Stripe integrations on the current, recommended APIs, with webhooks driving fulfillment and keys handle…

  9

  10## Current versions

  11

  12Latest Stripe API version: **2026-09-30.endive**. Use the latest API version and SDK unless the user specifie…

  13

  14| Language | Latest SDK |

  15|---|---|

  16| Ruby | 20.0.0 |
• 09

  AI SDK

  Fait coder l'agent avec l'AI SDK en vérifiant chaque API dans la doc embarquée de la version installée : streaming, outils, agents, useChat.+

  sf-ai-sdk/SKILL.md, de vercel/ai, licence Apache 2.0.

  1---

  2name: sf-ai-sdk

  3description: "Builds AI features with the AI SDK (the `ai` npm package): text generation, streaming, structur…

  4---

  5

  6# AI SDK

  7

  8Write AI SDK code that matches the version installed in the project. The SDK changes often (functions renamed…

  9

  10## What it is

  11

  12The AI SDK (`ai` on npm, from Vercel) is a TypeScript toolkit with one API across model providers for text ge…

  13

  14- Repository: https://github.com/vercel/ai

  15- Documentation: https://ai-sdk.dev/docs

  16
• 10

  MCP Server Builder

  Fait construire à l'agent un serveur MCP en TypeScript ou Python avec des outils bien nommés, sûrs et testés à l'Inspector puis par 10 questions d'évaluation.+

  sf-mcp-server/SKILL.md, de anthropics/skills, licence Apache 2.0.

  1---

  2name: sf-mcp-server

  3description: Builds high-quality MCP (Model Context Protocol) servers that let LLMs use an external API or se…

  4---

  5

  6# MCP Server Builder

  7

  8Build an MCP server whose tools an LLM can actually use to finish real tasks. Quality is measured by agent su…

  9

  10## When to use

  11

  12- Wrapping a REST or GraphQL API, a database, or a local tool as MCP tools.

  13- Adding tools, resources or prompts to an existing server.

  14- Reviewing a server's tool design, schemas and errors.

  15- Writing evaluations that prove an LLM can use the server.

  16

Extraits. Les dossiers complets sont livrés après le paiement.

Installation