Skip to content

Dianel555/DSkills / skills/grok-search

grok-search

Verified

Enhanced web search and real-time content retrieval via Grok API with forced tool routing. Use when: (1) Web search / information retrieval / fact-checking, (2) Webpage content extraction / URL parsing, (3) Breaking knowledge cutoff limits for current information, (4) Real-time news and technical documentation, (5) Multi-source information aggregation. Triggers: "search for", "find information about", "latest news", "current", "fetch webpage", "get content from URL". IMPORTANT: This skill REPLACES built-in WebSearch/WebFetch with Grok Search tools.

# GrokSearch CLI

Standalone command-line interface for Grok web search. No MCP dependency required.

## Installation

```bash
pip install httpx tenacity
```

## Layout

```text
groksearch_cli.py      # CLI entrypoint and compatibility facade
groksearch/           # Internal implementation modules
  cli.py              # argparse wiring
  commands.py         # command handlers
  config.py           # environment and persisted config
  http.py             # shared client and retry helpers
  provider.py         # Grok OpenAI-compatible provider
  tavily.py           # Tavily search/extract/map calls
  formatting.py       # JSON extraction and result merging
```

## Configuration

### Option 1: .env File (Recommended)

Create a `.env` file in the scripts directory:

```bash
cp .env.example .env
```

Edit `.env`:
```
GROK_API_URL=https://your-api-endpoint.com/v1
GROK_API_KEY=your-api-key-here
```

### Option 2: Environment Variables

```bash
export GROK_API_URL="https://your-api-endpoint.com/v1"
export GROK_API_KEY="your-api-key-here"
export TAVILY_API_KEY="your-tavily-key"  # optional
```

### Option 3: Command Line Arguments

```bash
python groksearch_cli.py --api-url "https://..." --api-key "sk-..." web_search -q "query"
```

## Commands

### web_search - Web Search

```bash
python groksearch_cli.py web_search --query "search terms" [options]

Options:
  -q, --query        Search query (required)
  -p, --platform     Focus platforms, e.g., "GitHub,Reddit"
  --min-results      Minimum results (default: 3)
  --max-results      Maximum results (default: 10)
  --extra-sources    Additional Tavily results to merge (default: 0)
  --raw              Output raw response without JSON parsing
```

Example:
```bash
python groksearch_cli.py web_search -q "latest Python 3.12 features" --max-results 5
```

### web_fetch - Fetch Webpage Content

```bash
python groksearch_cli.py web_fetch --url "https://..." [options]

Options:
  -u, --url          URL to fetch (required)
  -o, --out          Output file path (optional)
  --via              Fetch backend: grok|tavily (default: grok)
```

Example:
```bash
python groksearch_cli.py web_fetch -u "https://docs.python.org/3/whatsnew/3.12.html" -o python312.md
```

### web_map - Map Website Structure

```bash
python groksearch_cli.py web_map --url "https://..." [options]

Options:
  -u, --url          Root URL to map (required)
  --instructions     Natural language filter for crawler
  --max-depth        Max traversal depth (default: 1)
  --max-breadth      Max links per page (default: 20)
  --limit            Total link limit (default: 50)
  --timeout          Operation timeout in seconds (default: 150)
```

### get_config_info - Check Configuration

```bash
python groksearch_cli.py get_config_info [options]

Options:
  --no-test          Skip connection test
```

### switch_model - Switch Grok Model

```bash
python groksearch_cli.py switch_model --model "model-id"

Options:
  -m, --model        Model ID to switch to (required)
```

Example:
```bash
python groksearch_cli.py switch_model -m "grok-2-latest"
```

### toggle_builtin_tools - Toggle Built-in Tools

```bash
python groksearch_cli.py toggle_builtin_tools [options]

Options:
  -a, --action       Action: on/off/status (default: status)
  -r, --root         Project root path (default: auto-detect via .git)
```

Example:
```bash
# Disable built-in WebSearch/WebFetch
python groksearch_cli.py toggle_builtin_tools -a on

# Enable built-in tools
python groksearch_cli.py toggle_builtin_tools -a off

# Check status
python groksearch_cli.py toggle_builtin_tools -a status
```

## Output Format

- `web_search`: JSON array `[{title, url, description, provider?}]`
- `web_fetch`: Structured Markdown
- `web_map`: JSON object `{base_url, results, response_time}`
- `web_crawl`: JSON object `{base_url, results, response_time, usage?}`
- `web_research`: JSON object `{request_id, status, content, sources, usage?}`
- Other commands: JSON object

## .env File Search Order

1. Current working directory
2. Script directory (`scripts/`)
3. Parent directory of script

## Configuration Persistence

- Model settings: `~/.config/grok-search/config.json`
- Built-in tools toggle: `<project>/.claude/settings.json`

## Tavily Tuning

Search and extract are tunable via `.env` (or environment). All are optional; defaults follow
Tavily's own agent guidance.

| Variable | Default | Values | Notes |
|----------|---------|--------|-------|
| `TAVILY_SEARCH_DEPTH` | `advanced` | `advanced` / `basic` / `fast` / `ultra-fast` | `advanced` = 2 credits, highest relevance, reaches more sources — best for grounding niche/recent/multi-facet queries. Others = 1 credit. `ultra-fast` returns one summary per URL instead of reranked chunks. |
| `TAVILY_CHUNKS_PER_SOURCE` | `3` | `1`–`5` | Snippets (≤500 chars) per source; joined by `[...]`. More = stronger evidence per URL. |
| `TAVILY_TOPIC` | `general` | `general` / `news` / `finance` | `news` auto-adds `published_date`. |
| `TAVILY_TIME_RANGE` | *(unset)* | `day` / `week` / `month` / `year` | Publish-date window. Sources with no detectable date are kept unless filtered. |
| `TAVILY_INCLUDE_DOMAINS` | *(unset)* | comma-separated | Restrict or boost to these domains (max 300). |
| `TAVILY_EXCLUDE_DOMAINS` | *(unset)* | comma-separated | Drop these domains (max 150). |
| `TAVILY_INCLUDE_DOMAINS_MODE` | `filter` | `filter` / `boost` | `boost` also searches the wide web, so trusted sources are prioritized without risking empty results. Only sent when `TAVILY_INCLUDE_DOMAINS` is set. |
| `TAVILY_INCLUDE_ANSWER` | `false` | `false` / `basic` / `advanced` | LLM-generated answer. Extra cost; off by default. |
| `TAVILY_INCLUDE_USAGE` | `false` | `true` / `false` | Add credit `usage` to each response. Search attaches it as `tavily_usage` on the first merged result; map/crawl/research include a `usage` key in their JSON object. |
| `TAVILY_EXTRACT_DEPTH` | `basic` | `basic` / `advanced` | `advanced` handles tables, JS-rendered pages, structured data — higher latency and cost. Also used for crawl extraction. |
| `TAVILY_EXTRACT_TIMEOUT` | `30` | `1.0`–`60.0` | Tavily-side timeout for `web_fetch --via tavily`. The HTTP client waits 10s longer so Tavily's diagnosable error wins. Applies per attempt; with retries the total wall time can reach `timeout × attempts`. |
| `TAVILY_CRAWL_TIMEOUT` | `150` | `10`–`150` | Crawl-side timeout (API range). Client waits 10s longer. Applies per attempt; with retries the total wall time can reach `timeout × attempts`. |
| `TAVILY_CRAWL_MAX_DEPTH` | `1` | int | Crawl depth. |
| `TAVILY_CRAWL_MAX_BREADTH` | `20` | int | Links followed per page. |
| `TAVILY_CRAWL_LIMIT` | `50` | int | Total pages crawled. |
| `TAVILY_CRAWL_ALLOW_EXTERNAL` | `true` | `true` / `false` | `false` keeps the crawl on one site. |
| `TAVILY_CRAWL_SELECT_PATHS` | *(unset)* | comma-separated regexes | Include only matching paths. |
| `TAVILY_CRAWL_EXCLUDE_PATHS` | *(unset)* | comma-separated regexes | Exclude matching paths. |
| `TAVILY_RESEARCH_MODEL` | `auto` | `mini` / `pro` / `auto` | Research agent model. |
| `TAVILY_RESEARCH_CITATION_FORMAT` | `numbered` | `numbered` / `mla` / `apa` / `chicago` | Citation style in the report. |
| `TAVILY_RESEARCH_OUTPUT_LENGTH` | `standard` | `short` / `standard` / `long` | Report length. |
| `TAVILY_RESEARCH_OUTPUT_SCHEMA` | *(unset)* | path to a JSON file | Structured output. The file must be a JSON Schema with non-empty `properties`; each property needs `type` (`object`/`string`/`integer`/`number`/`array`) and `description`. Invalid schemas are reported before any request is sent. When set, `content` in the response is an object rather than a string. |
| `TAVILY_RESEARCH_TIMEOUT` | `300` | seconds | Total polling budget before returning `status: timeout` (with `request_id` so the task can be re-fetched). A poll HTTP/network failure returns `status: poll_error` with the same `request_id`. |
| `TAVILY_RESEARCH_POLL_INTERVAL` | `5` | seconds | Delay between status checks. |

Rate limits: 100 RPM (development key) / 1000 RPM (production). A `429` carries a
`retry-after` header, which the retry logic honors. `crawl` is capped at 100 RPM and
`research` at 20 RPM on both tiers.

Example — recent, trusted sources only:

```bash
TAVILY_TOPIC=news TAVILY_TIME_RANGE=week \
TAVILY_INCLUDE_DOMAINS="reuters.com,bloomberg.com" TAVILY_INCLUDE_DOMAINS_MODE=boost \
python groksearch_cli.py web_search -q "AI regulation" --extra-sources 5
```

Invalid values are reported per-key by `get_config_info` rather than silently ignored.

##  Acknowledgments

- Based on the original [GuDaStudio/GrokSearch](https://github.com/GuDaStudio/GrokSearch).
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 198kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 198kUpdated Claude CodeCodexVerifiedInstall