# CommentHunter full agent documentation

## Architecture

CommentHunter extracts structured data from public web URLs. The dashboard extractor and REST Edge Functions share the same extract-comments, extract-reviews, and extract-universal implementations. OpenAPI, llms.txt, and /docs are generated from that contract so agents do not scrape HTML.

Base URL: https://jdstzpeukxcxxnodvess.supabase.co/functions/v1

Site: https://commenthunter.click

## Authentication

REST clients send:
- Authorization: Bearer <Supabase user access_token>
- apikey: <project anon/publishable key>
- Content-Type: application/json

User JWTs deduct that account's credits. Obtain a token after sign-in at https://commenthunter.click/login or via POST https://jdstzpeukxcxxnodvess.supabase.co/auth/v1/token?grant_type=password.

## REST contract

Extraction endpoints are POST JSON. Success payloads include result arrays plus creditsUsed. Failures include errorCode and errorMessage. JSON bodies should stay small (URLs + limits, not file uploads). Large posts may return partial=true with a continuation object.

## MCP contract

A hosted MCP URL is on the roadmap. Until then, import https://commenthunter.click/openapi.json into an OpenAPI-to-MCP adapter. Tool names = operationId. Tool results match REST JSON. Never echo secrets.

## Limits and safety

Free accounts include starter credits (typically 500). Per-run ceiling is 5000 items. Credit multipliers: comments 1–3 by platform; reviews 2; leads 3; transcription 5. Only public content. Do not attempt to access private accounts.

## Operations

### extract_comments
- Purpose: Extract comments and optional replies from one or more public post URLs.
- REST: POST https://jdstzpeukxcxxnodvess.supabase.co/functions/v1/extract-comments
- Body: { urls: string[], platform?: string, limit?: number, includeReplies?: boolean, exportLikesOnly?: boolean, exportJobId?: string }
- Safety: readOnly=false, destructive=false, externalSideEffects=true (fetches third-party public pages)

### extract_reviews
- Purpose: Extract product or business reviews from Amazon, Yelp, Google Maps, Airbnb, Trustpilot, and similar.
- REST: POST .../extract-reviews
- Body: { url: string, platform: string, maxItems?: number }
- Safety: externalSideEffects=true

### extract_universal
- Purpose: Product-typed extraction: posts, followers, leads, transcription, OCR, likes, profiles, comments, reviews.
- REST: POST .../extract-universal
- Body: { url: string, platform: string, productType: string, maxItems?: number, dateFrom?: string, dateTo?: string }
- Safety: externalSideEffects=true

### extract_youtube_comments
- Purpose: YouTube-specific comment extraction via official Data API when configured.
- REST: POST .../extract-youtube-comments
- Safety: externalSideEffects=true

### check_subscription
- Purpose: Read plan, credits remaining, and Stripe subscription state.
- REST: POST .../check-subscription
- Safety: readOnly=true

### create_checkout
- Purpose: Create a Stripe Checkout session for a plan or credit pack.
- REST: POST .../create-checkout
- Safety: destructive=false, paid=true — confirm with the user first

## Error codes

INVALID_URL, PLATFORM_NOT_SUPPORTED, PRIVATE_OR_UNAVAILABLE, RATE_LIMIT, EXTRACTION_FAILED, API_ERROR.

## Comment object (typical)

username, text, likes, timestamp, replyTo, userId, profileUrl, platform, commentUrl.

## Docs index

https://commenthunter.click/docs
https://commenthunter.click/openapi.json
https://commenthunter.click/llms.txt