@johpaz/hive-sdk 0.1.3 → 0.1.5

package/packages/core/src/skills/bundled/web/web_research/SKILL.md

@@ -0,0 +1,119 @@
+---
+name: web_research
+description: "Search and synthesize information from multiple web sources into structured reports"
+version: 1.0.0
+author: Hive Team
+icon: "🔍"
+category: web
+permissions:
+  - internet_access
+dependencies: []
+tools: [web_search, web_fetch]
+
+# Structured skill fields
+triggers:
+  - "investigá sobre"
+  - "research"
+  - "buscá información de"
+  - "find information about"
+  - "qué es"
+  - "what is"
+  - "explicame"
+  - "explain"
+  - "últimos avances"
+  - "latest advances"
+  - "tendencias de"
+  - "trends in"
+  - "información actualizada"
+  - "current information"
+
+preferred_agents: []
+
+steps:
+  - step: 1
+    action: web_search
+    instruction: "Search for main topic using clear and concise query (max 6 words). Include year if relevant."
+    params:
+      query: "main topic keywords"
+      numResults: 8
+    output: raw_results
+
+  - step: 2
+    action: web_fetch
+    instruction: "Fetch the 2-3 most relevant URLs from step 1 to get full content"
+    params:
+      urls: "top 2-3 URLs from search"
+    output: detailed_content
+
+  - step: 3
+    action: web_search
+    instruction: "If gaps exist, do second targeted search with refined query"
+    params:
+      query: "refined query for gaps"
+    output: complementary_results
+
+  - step: 4
+    action: synthesize
+    instruction: "Combine all gathered information into structured response: summary, key points, sources"
+    output: final_report
+
+rules:
+  - "Always do at least 2 searches before concluding"
+  - "Never invent data — only use information actually found"
+  - "If no results found, notify user and suggest 2-3 alternative queries"
+  - "Cite all sources at end with full URLs"
+  - "Prioritize recent content (<1 year) unless historical context requested"
+  - "Cross-reference claims across multiple independent sources"
+
+output_format:
+  structure: markdown
+  sections:
+    - "summary"
+    - "key_points"
+    - "sources"
+  max_length: "500 words unless user requests more"
+
+examples:
+  - user_input: "investigá sobre los últimos avances en IA"
+    expected_behavior: "Search 'latest AI advances 2025' → fetch top 2-3 results → synthesize into structured report"
+
+  - user_input: "qué es un transformer en machine learning"
+    expected_behavior: "Search 'transformer machine learning explained' → fetch educational content → return clear summary"
+
+  - user_input: "buscá tendencias de quantum computing"
+    expected_behavior: "Search 'quantum computing trends 2025' → fetch authoritative sources → compile 5-7 key trends"
+---
+
+# Web Research Skill
+
+## Cuándo se Activa
+
+Esta skill se activa cuando el usuario necesita información actualizada de internet, verificar datos, o investigar temas específicos.
+
+## Herramientas Disponibles
+
+| Tool | Qué hace | Cuándo usarla |
+|------|----------|---------------|
+| `web_search` | Busca en internet, devuelve títulos, URLs, snippets | Búsqueda inicial, encontrar fuentes |
+| `web_fetch` | Descarga contenido completo de URL (HTML→Markdown) | Profundizar en resultados específicos |
+
+## Workflow
+
+1. **Búsqueda inicial** → `web_search({ query, numResults: 8 })`
+2. **Fetch contenido** → `web_fetch({ urls: top 2-3 })`
+3. **Búsqueda complementaria** → Segundo search si hay gaps
+4. **Síntesis** → summary + key points + sources
+
+## Mejores Prácticas
+
+- Queries específicos (máx 6 palabras)
+- Mínimo 2-3 fuentes independientes
+- Priorizar contenido reciente (<1 año)
+- Citas con URLs completas
+
+## Errores a Evitar
+
+- ❌ Inventar datos no encontrados
+- ❌ Concluir con una sola búsqueda
+- ❌ No verificar fecha de fuentes
+- ❌ Copiar contenido literal (usar paráfrasis)