feat: add Tavily as parallel search provider option

README.md

@@ -7,7 +7,9 @@ Search the web, get structured results, and extract page content as markdown —
 ## Quick Start
 
 1. **Install the skill** (pick one method below)
-2. **Set your API key**: `export QUERIT_API_KEY="your-key-here"`
+2. **Set your API key** (choose one provider):
+   - Querit: `export QUERIT_API_KEY="your-key-here"`
+   - Tavily: `export TAVILY_API_KEY="tvly-your-key-here"`
 3. **Use it**: Ask OpenClaw to "search the web for..." and it handles the rest
 
 ## Installation
@@ -31,17 +33,42 @@ Copy the skill files to `~/.openclaw/skills/querit-search/` and run `npm ci` in
 
 ## Configuration
 
-Get a free API key at [querit.ai](https://querit.ai) (1,000 queries/month, no credit card required).
+This skill supports two search providers. You only need one.
+
+### Querit (default)
 
-### Option A: Environment variable
+Get a free API key at [querit.ai](https://querit.ai) (1,000 queries/month, no credit card required).
 
 ```bash
 export QUERIT_API_KEY="your-key-here"
 ```
 
-Add to your `~/.bashrc`, `~/.zshrc`, or `~/.profile` to persist.
+### Tavily
+
+Get a free API key at [app.tavily.com](https://app.tavily.com) (1,000 credits/month, no credit card required).
+
+```bash
+export TAVILY_API_KEY="tvly-your-key-here"
+```
+
+### Provider Selection
+
+The provider is chosen automatically based on which API key is set. If both keys are present, Querit is used by default. Override with:
+
+```bash
+export SEARCH_PROVIDER="tavily"   # or "querit"
+```
+
+**Tavily-specific notes:**
+- Max 20 results per query (Querit supports up to 100)
+- `--lang` filter is not supported (Querit-only); a warning is printed if used
+- `--country` filter is supported
+- `--date` filter maps to Tavily time ranges: `d1`→day, `w1`→week, `m1`→month, `y1`→year
+- Content extraction (`content.js`) uses Tavily Extract with automatic fallback to the built-in jsdom pipeline
+
+### Additional config methods
 
-### Option B: OpenClaw config
+#### OpenClaw config
 
 In `~/.openclaw/openclaw.json`:
 
@@ -57,12 +84,13 @@ In `~/.openclaw/openclaw.json`:
 }
 ```
 
-### Option C: .env file
+#### .env file
 
 Create `~/.openclaw/.env`:
 
 ```
 QUERIT_API_KEY=your-key-here
+TAVILY_API_KEY=tvly-your-key-here
 ```
 
 ## Usage Examples

SKILL.md

@@ -4,7 +4,7 @@ description: >-
   Web search via Querit.ai API. Use when you need to search the web for
   documentation, current events, facts, or any web content. Returns
   structured results with titles, URLs, and snippets.
-metadata: {"openclaw":{"emoji":"🔎","requires":{"env":["QUERIT_API_KEY"]},"primaryEnv":"QUERIT_API_KEY","install":[{"id":"node","kind":"node","label":"Install npm dependencies"}]}}
+metadata: {"openclaw":{"emoji":"🔎","requires":{"env":[]},"optionalEnv":["QUERIT_API_KEY","TAVILY_API_KEY","SEARCH_PROVIDER"],"install":[{"id":"node","kind":"node","label":"Install npm dependencies"}]}}
 ---
 
 # Querit Search
@@ -13,7 +13,11 @@ Web search and content extraction via the Querit.ai API. No browser required.
 
 ## Setup
 
-Needs env: `QUERIT_API_KEY` — get a free key at https://querit.ai (1,000 queries/month).
+Needs one of:
+- `QUERIT_API_KEY` — get a free key at https://querit.ai (1,000 queries/month)
+- `TAVILY_API_KEY` — get a free key at https://app.tavily.com (1,000 credits/month)
+
+Optional: `SEARCH_PROVIDER` — set to `tavily` or `querit` to force a provider. If unset, auto-detects based on which API key is available.
 
 ## Search
 
@@ -87,7 +91,7 @@ Raw JSON array of result objects with fields: `url`, `title`, `snippet`, `page_a
 ## Limitations
 
 - Query limited to 72 characters (auto-truncated with warning)
-- Max 100 results per query
+- Max 100 results per query (Querit); max 20 results per query (Tavily)
 - Max 20 domains per site filter
 - Free tier: 1,000 queries/month, 1 QPS
 - Supported languages: english, japanese, korean, german, french, spanish, portuguese

content.js

@@ -17,7 +17,8 @@ function usage() {
 Fetches a URL and extracts the main content as readable markdown.
 
 Environment:
-  No API key needed — fetches pages directly.`);
+  TAVILY_API_KEY  Used if SEARCH_PROVIDER=tavily or only Tavily key is set
+  No API key needed for direct fetch (default).`);
 }
 
 function createTurndown() {
@@ -84,6 +85,54 @@ async function extractContent(url) {
   return lines.join("\n");
 }
 
+/**
+ * Determine whether to use Tavily Extract based on SEARCH_PROVIDER / API key presence.
+ */
+function useTavily() {
+  const explicit = process.env.SEARCH_PROVIDER;
+  if (explicit) return explicit.toLowerCase() === "tavily";
+  return !!process.env.TAVILY_API_KEY && !process.env.QUERIT_API_KEY;
+}
+
+/**
+ * Extract content using Tavily Extract API, falling back to jsdom pipeline on failure.
+ */
+async function tavilyExtract(url) {
+  const { tavily } = await import("@tavily/core");
+  const client = tavily({ apiKey: process.env.TAVILY_API_KEY });
+  const response = await client.extract([url], { format: "markdown" });
+
+  if (!response.results || response.results.length === 0) {
+    throw new Error("Tavily Extract returned no results");
+  }
+
+  const result = response.results[0];
+  return result.raw_content || "(no content extracted)";
+}
+
+/**
+ * Top-level extraction dispatcher: tries Tavily if configured, falls back to jsdom.
+ */
+async function extractWithProvider(url) {
+  const explicit = process.env.SEARCH_PROVIDER;
+  if (explicit && explicit.toLowerCase() === "tavily" && !process.env.TAVILY_API_KEY) {
+    console.error(
+      "Warning: SEARCH_PROVIDER=tavily but TAVILY_API_KEY is not set; falling back to direct fetch"
+    );
+  }
+
+  if (useTavily() && process.env.TAVILY_API_KEY) {
+    try {
+      return await tavilyExtract(url);
+    } catch (err) {
+      console.error(
+        `Warning: Tavily Extract failed (${err.message}), falling back to direct fetch`
+      );
+    }
+  }
+  return extractContent(url);
+}
+
 async function main() {
   const url = process.argv[2];
 
@@ -101,7 +150,7 @@ async function main() {
   }
 
   try {
-    const markdown = await extractContent(url);
+    const markdown = await extractWithProvider(url);
     console.log(markdown);
   } catch (err) {
     console.error(`Error extracting content from ${url}: ${err.message}`);