researching-features

majiayu000's avatarfrom majiayu000

Use this whenever a user wants to add a new feature or explitly states to research a feature/API or building a plan for a new feature. It itnerviews the user for feature details (if not provided), research the best API/service for their needs, confirm choice, then gather all implementation notes for their request and save them as a .claude/plans file.

⭐0stars🔀0forks📁View on GitHub🕐Updated Jan 5, 2026
[API Helpers]

When & Why to Use This Skill

The Feature Researcher skill is a powerful developer tool designed to streamline the technical discovery and planning phase of software development. It automates the process of identifying the best APIs and services for specific project requirements, conducting deep-dive research into documentation, and generating structured implementation roadmaps. By bridging the gap between a feature concept and a technical plan, it ensures developers have all necessary authentication steps, endpoints, and usage patterns organized in a dedicated workspace file before coding begins.

Use Cases

  • Technical Discovery: Automatically researching and comparing third-party providers (e.g., Stripe vs. PayPal for payments) based on specific project constraints and budget.
  • Implementation Planning: Generating comprehensive .md files in .claude/plans/ that outline authentication flows, core API methods, and error handling strategies for new features.
  • Requirement Refinement: Using an interactive interview process to clarify vague feature requests and transform them into actionable technical specifications.
  • Documentation Consolidation: Gathering official API documentation, rate limits, and pricing details into a single, project-specific context file for easier reference during development.
name"Researching Features"
description"Use this whenever a user wants to add a new feature or explitly states to research a feature/API or building a plan for a new feature. It itnerviews the user for feature details (if not provided), research the best API/service for their needs, confirm choice, then gather all implementation notes for their request and save them as a .claude/plans file. "
version"1.0.0"
dependencies["context7", "mcp-api", "python>=3.8"]
allowed-tools["context7", "mcp", "file_write"]

Feature Researcher

Instructions

When requested to research a feature:

  1. User Interview

    • If the user's requirements are unclear, politely ask for more details (deatails on feature, free/paid API options, constraints).
    • If details are provided, proceed directly.

  2. Service & API Discovery

    • Take the user's answers and consider them in your search
    • You MUST use context7 to identify the APIs/services/libraries that best match the user's requirements.
    • DO NOT use web_search - tell teh user to get context7
    • Only use web_search if the the user doesnt want to use context7
    • Go with the top 3 options that the tools return/suggest

  3. User Confirmation

    • Summarize every provider you found and suggest
    • After selecting the best API/service, briefly summarize your choice and reasons.
    • Ask the user to confirm before proceeding with implementation research.

  4. Implementation Notes Gathering

    • VERY IMPORTANT: Before you start, look at the spec-sheet.md for the specs for this projects and claude.md to understand the context - this way you know the tech stack to a build a plan for.
    • Once confirmed, use Context7 to retrieve official docs, key endpoints, authentication steps, usage patterns, and constraints for the selected API/service.
    • Structure your notes clearly around:
      • Have page and UI elements to be built first before backend functions etc
      • Authentication
      • Setup and Initialization
      • Core Endpoints/Methods
      • Example Requests/Responses d
      • Error Handling
      • Rate Limits or Pricing

  5. Save Implementation Plan

    • Compile all notes and implementation steps into a .md file.
    • Create a plan in .claude/plans/plan-[feature-name].md.
    • Notify the user where to find their plan.

Examples

  • Input: "I want live chat in my app. What service is best?" Output:

    1. Interview user for scale, preferred integrations.
    2. Research providers (Twilio Conversations, Sendbird, CometChat).
    3. Suggest Sendbird based on docs and usage.
    4. After user approval, gather usage notes, endpoints, sample code.
    5. Save results to .claude/plans/plan-feature-live-chat.md.

  • Input: "Add online payments (API/service of your choice)" Output:
    Same flow, ending with a plan file like .claude/plans/plan-feature-payments.md