@apmantza/greedysearch-pi 1.7.6 → 1.8.0

package/skills/greedy-search/skill.md

@@ -1,44 +1,44 @@
----
-name: greedy-search
-description: Live web search via Perplexity, Bing, and Google AI in parallel. Use for library docs, recent framework changes, error messages, dependency selection, or anything where training data may be stale. NOT for codebase search.
----
-
-# GreedySearch — Live Web Search
-
-Runs Perplexity, Bing Copilot, and Google AI in parallel. Gemini synthesizes results.
-
-## greedy_search
-
-```
-greedy_search({ query: "React 19 changes", depth: "standard" })
-```
-
-| Parameter | Type | Default | Description |
-|-----------|------|---------|-------------|
-| `query` | string | required | Search question |
-| `engine` | string | `"all"` | `all`, `perplexity`, `bing`, `google`, `gemini` |
-| `depth` | string | `"standard"` | `fast`, `standard`, `deep` |
-| `fullAnswer` | boolean | `false` | Full answer vs ~300 char summary |
-
-| Depth | Engines | Synthesis | Source Fetch | Time |
-|-------|---------|-----------|--------------|------|
-| `fast` | 1 | — | — | 15-30s |
-| `standard` | 3 | Gemini | — | 30-90s |
-| `deep` | 3 | Gemini | top 5 | 60-180s |
-
-**When engines agree** → high confidence. **When they diverge** → note both perspectives.
-
-## coding_task
-
-Second opinion from Gemini/Copilot on hard problems.
-
-```
-coding_task({ task: "debug race condition", mode: "debug", engine: "gemini" })
-```
-
-| Parameter | Type | Default | Options |
-|-----------|------|---------|---------|
-| `task` | string | required | — |
-| `engine` | string | `"gemini"` | `gemini`, `copilot`, `all` |
-| `mode` | string | `"code"` | `debug`, `plan`, `review`, `test`, `code` |
-| `context` | string | — | Code snippet |
+---
+name: greedy-search
+description: Live web search via Perplexity, Bing, and Google AI in parallel. Use for library docs, recent framework changes, error messages, dependency selection, or anything where training data may be stale. NOT for codebase search.
+---
+
+# GreedySearch — Live Web Search
+
+Runs Perplexity, Bing Copilot, and Google AI in parallel. Gemini synthesizes results.
+
+## greedy_search
+
+```
+greedy_search({ query: "React 19 changes", depth: "standard" })
+```
+
+| Parameter | Type | Default | Description |
+|-----------|------|---------|-------------|
+| `query` | string | required | Search question |
+| `engine` | string | `"all"` | `all`, `perplexity`, `bing`, `google`, `gemini` |
+| `depth` | string | `"standard"` | `fast`, `standard`, `deep` |
+| `fullAnswer` | boolean | `false` | Full answer vs ~300 char summary |
+
+| Depth | Engines | Synthesis | Source Fetch | Time |
+|-------|---------|-----------|--------------|------|
+| `fast` | 1 | — | — | 15-30s |
+| `standard` | 3 | Gemini | — | 30-90s |
+| `deep` | 3 | Gemini | top 5 | 60-180s |
+
+**When engines agree** → high confidence. **When they diverge** → note both perspectives.
+
+## coding_task
+
+Second opinion from Gemini/Copilot on hard problems.
+
+```
+coding_task({ task: "debug race condition", mode: "debug", engine: "gemini" })
+```
+
+| Parameter | Type | Default | Options |
+|-----------|------|---------|---------|
+| `task` | string | required | — |
+| `engine` | string | `"gemini"` | `gemini`, `copilot`, `all` |
+| `mode` | string | `"code"` | `debug`, `plan`, `review`, `test`, `code` |
+| `context` | string | — | Code snippet |

package/src/fetcher.mjs

@@ -178,6 +178,7 @@ export async function fetchSourceHttp(url, options = {}) {
 
 		const contentType = response.headers.get("content-type") || "";
 		const finalUrl = response.url;
+		const lastModified = response.headers.get("last-modified") || "";
 
 		// Handle raw text/plain from GitHub (raw file content)
 		if (
@@ -191,6 +192,11 @@ export async function fetchSourceHttp(url, options = {}) {
 				finalUrl,
 				status: response.status,
 				title: finalUrl.split("/").pop() || "GitHub File",
+				byline: "",
+				siteName: "GitHub",
+				lang: "",
+				publishedTime: lastModified,
+				lastModified,
 				markdown: text,
 				contentLength: text.length,
 				excerpt: text.slice(0, 300).replace(/\n/g, " "),
@@ -250,6 +256,11 @@ export async function fetchSourceHttp(url, options = {}) {
 			finalUrl,
 			status: response.status,
 			title: extracted.title,
+			byline: extracted.byline,
+			siteName: extracted.siteName,
+			lang: extracted.lang,
+			publishedTime: extracted.publishedTime || lastModified,
+			lastModified,
 			markdown: extracted.markdown,
 			excerpt: extracted.excerpt,
 			contentLength: extracted.markdown.length,
@@ -437,6 +448,29 @@ function isNetworkErrorRetryableWithBrowser(error) {
 	);
 }
 
+/**
+ * Extract a date string from <meta> tags (Open Graph, schema.org, standard)
+ * Returns ISO string or empty string.
+ */
+function extractMetaDate(document) {
+	const selectors = [
+		'meta[property="article:published_time"]',
+		'meta[name="article:published_time"]',
+		'meta[property="og:published_time"]',
+		'meta[name="publication_date"]',
+		'meta[name="date"]',
+		'meta[itemprop="datePublished"]',
+		'time[itemprop="datePublished"]',
+		'meta[name="DC.date"]',
+	];
+	for (const sel of selectors) {
+		const el = document.querySelector(sel);
+		const val = el?.getAttribute("content") || el?.getAttribute("datetime") || "";
+		if (val) return val;
+	}
+	return "";
+}
+
 /**
  * Extract readable content using Mozilla Readability + Turndown
  */
@@ -452,8 +486,14 @@ function extractContent(html, url) {
 		const markdown = turndown.turndown(article.content);
 		const cleanMarkdown = markdown.replace(/\n{3,}/g, "\n\n").trim();
 
+		const publishedTime = article.publishedTime || extractMetaDate(document) || "";
+
 		return {
 			title: article.title || document.title || url,
+			byline: article.byline || "",
+			siteName: article.siteName || "",
+			lang: article.lang || "",
+			publishedTime,
 			markdown: cleanMarkdown,
 			excerpt: cleanMarkdown.slice(0, 300).replace(/\n/g, " "),
 		};
@@ -472,6 +512,10 @@ function extractContent(html, url) {
 
 		return {
 			title: document.title || url,
+			byline: "",
+			siteName: "",
+			lang: "",
+			publishedTime: extractMetaDate(document),
 			markdown: cleanText,
 			excerpt: cleanText.slice(0, 300),
 		};
@@ -480,6 +524,10 @@ function extractContent(html, url) {
 	// Last resort
 	return {
 		title: url,
+		byline: "",
+		siteName: "",
+		lang: "",
+		publishedTime: "",
 		markdown: "",
 		excerpt: "",
 	};