web-research

Category: Tools & Productivity | Uploader: mrvarmazyar | Downloads: 0 | Version: v1.0（Latest）

Use for all web searches and URL fetching — unified CLI combining tinyfish search + Groq summarization. Replaces both native WebSearch and native WebFetch. Triggers on: research tasks, finding docs, "search for X", "look up X", fetching any public URL. Skip for GitHub repos/issues (use gh CLI), Jira/Confluence (use MCP tools), internal/authenticated URLs.

Changelog: Source: GitHub https://github.com/mrvarmazyar/web-research

Directory Structure

Current level: tree/main/

📁 .github/
- 📁 workflows/
  - 📄 ci.yml 434 B
  - 📄 release.yml 5.6 KB
📁 cmd/
- 📁 wr/
  - 📄 main.go 9.9 KB
  - 📄 options.go 2.4 KB
  - 📄 options_test.go 3.1 KB
- 📁 wr-mcp/
  - 📄 format.go 492 B
  - 📄 main.go 4.0 KB
📁 docs/
- 📄 mcp.md 2.6 KB
📁 internal/
- 📁 cache/
  - 📄 cache.go 991 B
- 📁 fetch/
  - 📄 fetch.go 1.3 KB
- 📁 research/
  - 📄 service.go 5.6 KB
  - 📄 types.go 2.2 KB
  - 📄 validation.go 1.5 KB
  - 📄 validation_test.go 4.4 KB
- 📁 retrieve/
  - 📄 chunk.go 340 B
  - 📄 chunk_test.go 1.2 KB
  - 📄 retrieve.go 1.4 KB
  - 📄 retrieve_test.go 2.0 KB
  - 📄 tfidf.go 2.2 KB
  - 📄 tfidf_test.go 1.5 KB
- 📁 search/
  - 📄 tinyfish.go 1.2 KB
- 📁 summarize/
  - 📄 copilot.go 1.6 KB
  - 📄 copilot_test.go 2.9 KB
  - 📄 groq.go 1.7 KB
  - 📄 options.go 2.8 KB
  - 📄 options_test.go 1.5 KB
📄 .gitignore 24 B
📄 go.mod 609 B
📄 go.sum 8.3 KB
📄 README.md 8.2 KB
📄 SKILL.md 1.8 KB

SKILL.md

---
name: web-research
description: Use for all web searches and URL fetching — unified CLI combining tinyfish search + Groq summarization. Replaces both native WebSearch and native WebFetch. Triggers on: research tasks, finding docs, "search for X", "look up X", fetching any public URL. Skip for GitHub repos/issues (use gh CLI), Jira/Confluence (use MCP tools), internal/authenticated URLs.
---

# Web Research (`wr`)

Unified web research CLI. Tinyfish for search, Groq for summarization, local cache for speed.

## Auto-Setup Check

```bash
which wr || { echo "Install: cd ~/web-research && go build -o ~/.local/bin/wr ./cmd/wr"; }
wr setup
```

## Commands

```bash
wr search "<query>"              # search only → URLs + snippets
wr fetch "<url>" "<prompt>"      # fetch URL → Groq summary
wr research "<query>"            # search + fetch top 3 + summarize
wr setup                         # check TINYFISH_API_KEY + GROQ_API_KEY
```

## Optimal Usage Pattern

```
1. wr research "query"           → full pipeline, best for research tasks
2. wr search "query"             → when you just need URLs first
   wr fetch "<url>" "<prompt>"   → then fetch specific result
```

## When to Skip

- GitHub issues/PRs/repos → `gh` CLI
- Jira/Confluence → Atlassian MCP tools  
- Internal/authenticated URLs → native WebFetch
- Already have exact answer → no need to search

## Env Vars

```bash
export TINYFISH_API_KEY="..."   # https://agent.tinyfish.ai → API Keys
export GROQ_API_KEY="..."       # https://console.groq.com → API Keys (free)
export WR_CACHE_DAYS=7          # optional, default 7
```

## Fallback

If `wr` not installed or `wr setup` shows missing keys:
- Search → fall back to native WebSearch
- Fetch → fall back to `~/.claude/skills/web-fetch-local/fetch.sh`

Comments 0

Latest | Hot

Please login before commenting.

Loading comments...

web-research

Directory Structure

SKILL.md

Report

Notice