Skip to main content

Documentation Index

Fetch the complete documentation index at: https://usenaive.ai/docs/llms.txt

Use this file to discover all available pages before exploring further.

Overview

Three search modes, each with increasing depth and cost:
CommandDescriptionCost
naive search <query>Quick web search1 credit
naive search url <url>Extract content from a URL1 credit
naive search research <query>Deep multi-source AI research (async)3–8 credits

naive search "best project management tools 2026"
naive search "competitor analysis SaaS pricing" --count 10

Options

FlagRequiredDescription
queryYesSearch query (positional argument)
--count <n>NoNumber of results (default: 5, max: 20)

Output

{
  "success": true,
  "action": "search",
  "result": {
    "results": [
      { "title": "Top PM Tools in 2026", "url": "https://example.com/article", "snippet": "A comprehensive look at...", "score": 0.95 },
      { "title": "Building Better Teams", "url": "https://blog.example.com/post", "snippet": "How to organize..." }
    ],
    "credits_used": 1,
    "credits_remaining": 99,
    "provider": "web"
  },
  "next_steps": [
    { "command": "naive search url https://example.com/article", "description": "Extract full content from the top result" },
    { "command": "naive search research \"best project management tools 2026\"", "description": "Run deep research on this topic" },
    { "command": "naive search \"best project management tools 2026\" --count 10", "description": "Get more results" }
  ],
  "hints": ["Found 5 results", "Use 'naive search url <url>' to extract full page content", "Cost: 1 credit"]
}

URL Extraction

Fetches a URL and extracts content. Optionally uses AI to extract specific information.
naive search url https://example.com/pricing
naive search url https://news.ycombinator.com --extract "top 5 stories with links"
naive search url https://docs.example.com/api --extract "authentication methods"

Options

FlagRequiredDescription
urlYesURL to fetch and extract from
--extract <prompt>NoAI extraction prompt for targeted information

Output

{
  "success": true,
  "action": "search.url",
  "result": {
    "url": "https://example.com/pricing",
    "title": "",
    "content": "...(raw text content, up to 5000 chars)...",
    "extracted": "Free: $0/mo, Pro: $29/mo, Enterprise: custom",
    "credits_used": 1,
    "credits_remaining": 98
  },
  "next_steps": [
    { "command": "naive search url https://example.com/pricing --extract \"specific question\"", "description": "Re-extract with a different focus" },
    { "command": "naive search \"related topic\"", "description": "Search for related information" }
  ],
  "hints": ["Use --extract to focus on specific information from the page", "Cost: 1 credit"]
}
When --extract is provided and AI extraction is configured, an LLM extracts the specific information you requested from the page content. Without --extract, you get the raw text content.

Deep Research

Multi-source, AI-synthesized research. Runs synchronously for “quick” depth, asynchronously for thorough/exhaustive.
naive search research "market size for AI developer tools" --depth thorough
naive search research "competitive analysis: Figma vs Canva" --depth exhaustive --wait
naive search research "microservice auth best practices" --depth quick

Options

FlagRequiredDescription
queryYesResearch topic
--depth <level>Noquick (default), thorough, or exhaustive
--waitNoBlock until research completes (for thorough/exhaustive)

Depth Levels

LevelSourcesBehaviorCost
quick5Synchronous (immediate response)3 credits
thorough8Async job (~45s)5 credits
exhaustive15Async job (~90s)8 credits

Output (quick — inline response)

{
  "success": true,
  "action": "search.research",
  "result": {
    "synthesis": "Based on analysis of 5 sources...",
    "sources": [
      { "url": "https://...", "title": "...", "relevance": 0.9 }
    ],
    "credits_used": 3,
    "credits_remaining": 97
  },
  "next_steps": [
    { "command": "naive search \"follow-up query\"", "description": "Quick search for additional context" }
  ],
  "hints": ["Research completed synchronously (quick depth)"]
}

Output (thorough/exhaustive — submitted)

{
  "success": true,
  "action": "search.research.submitted",
  "result": { "job_id": "job-uuid-123", "status": "queued", "depth": "thorough" },
  "next_steps": [
    { "command": "naive jobs get job-uuid-123", "description": "Check research progress" },
    { "command": "naive jobs --status running", "description": "List all running jobs" }
  ],
  "hints": [
    "Research job submitted (depth: thorough)",
    "Credits will be deducted only on successful completion",
    "Expected completion: ~45s"
  ]
}

Output (completed — with --wait)

{
  "success": true,
  "action": "search.research.completed",
  "result": {
    "id": "job-uuid-123",
    "status": "completed",
    "type": "deep_research",
    "result": {
      "synthesis": "Based on analysis of 8 sources...",
      "sources": [
        { "url": "https://...", "title": "...", "relevance": 0.92 }
      ]
    },
    "credits_used": 5
  },
  "next_steps": [
    { "command": "naive search \"related topic\"", "description": "Search for additional information" },
    { "command": "naive search url <url>", "description": "Deep-dive into a specific source from the report" }
  ],
  "hints": ["Research completed successfully"]
}
Credit behavior: Credits are NOT deducted at submission. They are deducted only when the job completes successfully. Failed or cancelled research jobs cost nothing.