michealwayne/fe-tools-frontend-resource-finder

Frontend Resource Finder

Goal

Recommend the most relevant resources from this repository's curated indexes instead of dumping raw search results.

Source Selection

1. Detect the user's input language.
2. Start from the repository root one level above this package: ../README.md and ../README-en.md.
3. If the request is mainly English, read ../README-en.md.
4. Otherwise, read ../README.md.
5. If the preferred file is missing the needed category, check the sibling README as a fallback.
6. Do not treat package-level READMEs in utils/ as the curated resource index unless the root README pair is unavailable.

Workflow

1. Classify the question first. Prefer one primary category and at most one secondary category:
   • Standards and docs: HTML, CSS, JavaScript, TypeScript, Web APIs, RFC, WASM
   • Compatibility and browser support
   • CSS tooling and styling systems
   • JS/TS libraries and data tooling
   • Request, storage, cache, offline, performance
   • React, Vue, Node.js, build tooling
   • Testing, security, encryption
   • AI tools, coding assistants, agent platforms, agent skills
   • Cross-platform, hybrid, mini-program, desktop
2. Verify the target README exists before relying on it.
3. Search only the matching README sections instead of treating the README as full-text search only.
4. Pick 3-5 resources, not more, unless the user explicitly asks for a long list.
5. For each resource, explain why it matches the question:
   • What problem it solves
   • Why it is a better fit than adjacent options in this directory
   • How it fits modern frontend or AI coding workflows when relevant
6. Prioritize resources that are:
   • Official docs or de facto standards for learning and correctness
   • Practical and current for modern frontend stacks
   • Helpful for AI-assisted development, automation, or agent workflows when the question touches those areas
7. Avoid recommending obviously outdated items unless the user is asking about legacy support or historical context.

Output Format

Use this structure:

1. Category: one short label for how you classified the question.
2. Recommended resources: 3-5 items.
3. For each item include:
   • Resource name
   • Short reason tied to the user's scenario
   • A note about modern frontend / AI coding relevance when useful
4. Why these over others: one short comparison sentence or paragraph.
5. Next step: suggest what to read or try first.

Heuristics

• For broad "how do I learn X" questions, include at least:
  • one official reference
  • one practical ecosystem tool or library
  • one productivity or AI-oriented resource when it meaningfully helps
• For library selection questions, bias toward:
  • smaller, actively useful tools for modern TS/frontend workflows
  • ecosystem fit over generic popularity
• For AI coding questions:
  • combine model/tool platforms with frontend-specific references when possible
  • if the user is building agents, include agent skills or agent platform entries from the README when relevant
• For compatibility questions:
  • lead with compatibility databases, then the relevant official spec/doc
• For legacy-browser or low-version environments:
  • it is acceptable to recommend older resources if they are specifically relevant to that constraint

Search Hints

• Use README headings and nearby table rows instead of scanning the whole file every time.
• Common anchors to inspect:
  • 1.2 compatibility
  • 1.3 CSS tools
  • 1.4 JS plugins/libraries
  • 1.5 Vue
  • 1.6 React
  • 1.7 NodeJS and build
  • 1.10 testing/security/encryption
  • 1.11 AI
  • 1.13 IDE plugins
• When the query mentions skills, agents, copilots, or AI coding workflows, inspect the AI and agent-related parts first.
• When the query is about project starters or scaffolds, also inspect ../project-templates/README.md before answering.

Constraints

• Do not answer with a plain keyword match list.
• Do not recommend more than 5 resources by default.
• Do not fabricate resources that are not in the selected README file or ../project-templates/README.md when template-related.
• Keep the answer decision-oriented and scenario-aware.