Kagi FastGPT

Get AI-generated answers with cited web sources using Kagi's FastGPT API. FastGPT runs a full web search under the hood and synthesizes results into a concise answer — ideal for factual questions, API lookups, and current-events queries.

This skill uses a Go binary for fast startup and no runtime dependencies. The binary can be downloaded pre-built or compiled from source.

Setup

Requires a Kagi account with API access enabled. Uses the same KAGI_API_KEY as the kagi-search skill.

1. Create an account at https://kagi.com/signup
2. Navigate to Settings → Advanced → API portal: https://kagi.com/settings/api
3. Generate an API Token
4. Add funds at: https://kagi.com/settings/billing_api
5. Add to your shell profile (~/.profile or ~/.zprofile):

   export KAGI_API_KEY="your-api-key-here"
   

6. Install the binary — see Installation below

Pricing

1.5¢ per query ($15 USD per 1000 queries). Cached responses are free.

Usage

{baseDir}/kagi-fastgpt.sh "query"                        # Ask a question (default)
{baseDir}/kagi-fastgpt.sh "query" --json                 # JSON output
{baseDir}/kagi-fastgpt.sh "query" --no-refs              # Answer only, no references
{baseDir}/kagi-fastgpt.sh "query" --no-cache             # Bypass response cache
{baseDir}/kagi-fastgpt.sh "query" --timeout 60           # Custom timeout (default: 30s)

Options

Output

Default (text)

Prints the synthesized answer, followed by a numbered reference list:

Python 3.11 was released on October 24, 2022 and introduced several improvements...

--- References ---
[1] What's New In Python 3.11 — Python 3.11.3 documentation
    https://docs.python.org/3/whatsnew/3.11.html
    The headline changes in Python 3.11 include significant performance improvements...
[2] ...

Token usage and API balance are printed to stderr.

JSON (--json)

Returns a JSON object with:

• query — the original query
• output — the synthesized answer
• tokens — tokens consumed
• references[] — array of { title, url, snippet } objects
• meta — API metadata (id, node, ms)

When to Use

• Use kagi-fastgpt when you need a direct answer synthesized from web sources (e.g. "What version of X was released last month?", "How do I configure Y?")
• Use kagi-search when you need raw search results to scan, compare, or extract data from yourself
• Use web-browser when you need to interact with a page or the content is behind JavaScript

Installation

Option A — Download pre-built binary (no Go required)

OS=$(uname -s | tr '[:upper:]' '[:lower:]')
ARCH=$(uname -m)
case "$ARCH" in
  x86_64)        ARCH="amd64" ;;
  aarch64|arm64) ARCH="arm64" ;;
esac

TAG=$(curl -fsSL "https://api.github.com/repos/joelazar/kagi-skills/releases/latest" | grep '"tag_name"' | cut -d'"' -f4)
BINARY="kagi-fastgpt_${TAG}_${OS}_${ARCH}"

mkdir -p {baseDir}/.bin
curl -fsSL "https://github.com/joelazar/kagi-skills/releases/download/${TAG}/${BINARY}" \
  -o {baseDir}/.bin/kagi-fastgpt
chmod +x {baseDir}/.bin/kagi-fastgpt

# Verify checksum (recommended)
curl -fsSL "https://github.com/joelazar/kagi-skills/releases/download/${TAG}/checksums.txt" | \
  grep "${BINARY}" | sha256sum --check

Pre-built binaries are available for Linux and macOS (amd64 + arm64) and Windows (amd64).

Option B — Build from source (requires Go 1.26+)

cd {baseDir} && go build -o .bin/kagi-fastgpt .

Alternatively, just run {baseDir}/kagi-fastgpt.sh directly — the wrapper auto-builds on first run if Go is available.

The binary has no external dependencies — only the Go standard library.