Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.bytestack.com/llms.txt

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

ByteStack lets you query the open web the same way you’d ask a question to a colleague — in plain English. You write a prompt describing what you want to find, and ByteStack executes it across thousands of web sources, returning structured records you can explore, export, or pipe into downstream workflows. No query language, no manual scraping, no API wrangling required.

​
How queries work

When you submit a query, ByteStack interprets your prompt, identifies the most relevant sources, fetches matching content, and returns structured results. You can optionally narrow the query to specific platforms, set a date range, or cap the number of records returned.
1

Write your prompt

Describe what you want to find in plain English. Be as specific as possible — mention platforms, time ranges, and the subject you care about.
2

ByteStack executes the query

ByteStack routes your prompt to the relevant sources, scrapes matching content, and runs any requested analysis (such as sentiment classification).
3

Review structured results

Results appear in your project workspace as structured records with source, author, content, sentiment, and timestamp fields.

​
Query parameters

Every query accepts the following parameters:
ParameterTypeDescription
promptstringYour natural language query (required)
sourcesarrayPlatforms to query. Use ["all"] to search all sources
date_rangeobjectRestrict results to a time window (from and to as ISO 8601 dates)
limitintegerMaximum number of records to return (default: 100)

​
Example queries

The following examples show how to structure a query via the API: 
{
  "prompt": "How many people mentioned Acme on X in the last week?",
  "sources": ["x"],
  "date_range": {
    "from": "2026-04-23",
    "to": "2026-04-30"
  },
  "limit": 500
}

​
Result structure

Each query returns an array of records. Here is an example response: 
{
  "query_id": "qry_7f3a2c01",
  "total": 3,
  "records": [
    {
      "id": "3a8f2c01",
      "source": "x",
      "author": "jakemorrrison",
      "content": "Acme's new product is honestly the best thing I've used this year.",
      "sentiment": "positive",
      "timestamp": "2026-04-28T09:14:22Z",
      "url": "https://x.com/jakemorrrison/status/..."
    },
    {
      "id": "7b2e9d44",
      "source": "reddit",
      "author": "priya_p",
      "content": "Not impressed with Acme support — took 3 days to get a response.",
      "sentiment": "negative",
      "timestamp": "2026-04-27T16:45:10Z",
      "url": "https://reddit.com/r/startups/comments/..."
    },
    {
      "id": "e1c04af7",
      "source": "linkedin",
      "author": "Marcus Chen",
      "content": "Partnering with Acme has been a solid move for our pipeline.",
      "sentiment": "positive",
      "timestamp": "2026-04-26T11:30:00Z",
      "url": "https://linkedin.com/posts/..."
    }
  ]
}
Each record includes the following fields:
  • id — unique record identifier
  • source — the platform the record was fetched from
  • author — username or display name of the content creator
  • content — the raw text of the post or mention
  • sentiment — positive, negative, or neutral
  • timestamp — ISO 8601 datetime of when the content was published
  • url — direct link to the original content
Add sentiment analysis to any query by including the phrase “classify by sentiment” or “what is the sentiment” in your prompt. ByteStack will automatically run its sentiment model and populate the sentiment field on each record.

​
Prompt writing FAQ

The most effective prompts are specific about three things: subject (what or who you’re researching), platform (where you want to look), and time frame (how recent the data should be).A vague prompt like “what do people say about my brand” will return broad, unfocused results. A specific prompt like “What are the top complaints about Acme’s pricing on Reddit and X in the last 30 days?” gives ByteStack a clear scope and returns actionable data.You don’t need to use structured query syntax — write naturally, as if briefing a researcher.
Yes. Within your project workspace, you can refine a query or ask follow-up questions against the same result set without re-running a full fetch. Use the prompt bar in the dashboard to continue the conversation.
Date range availability varies by source. Most social platforms support queries going back 12–24 months. Use the date_range parameter to target a specific window and avoid fetching more data than you need.
ByteStack returns an empty array for that source and continues fetching from others. The response includes a sources_queried summary so you can see which platforms were searched and how many records each returned.