scavio-g2

Search G2, the B2B software review site, read a full product profile with pricing and features, and pull faceted reviews with exact per-star counts. 3 endpoints, 5 credits each.

G2 via Scavio

Search G2, the B2B software review site, open any product's full profile with its pricing editions, features and alternatives, and pull reviews with exact per-star counts and company-size / role / industry / region facets. All three endpoints return structured JSON.

When to trigger

Use this skill when the user asks to:

• Find B2B software products in a category on G2
• Read a software product's G2 profile: rating histogram, pricing editions, features, integrations, alternatives
• Mine G2 reviews for what enterprise buyers, admins or a specific region actually say
• Compare a product against its listed alternatives and head-to-head comparisons
• Do competitive intelligence, win/loss research, or voice-of-customer analysis on SaaS

Cost: 5 credits per call

G2 is 5 credits per request - the most expensive platform on Scavio, and the only one at this tier. g2.com bills a premium upstream on every single fetch, with no cheap mode available. Three calls is 15 credits.

Plan the run before making it, tell the user the credit cost up front, and never loop over search results calling /product on each one.

Setup

Get a free API key at https://scavio.dev (50 free credits to get started, no card required):

bash

export SCAVIO_API_KEY=sk_live_your_key

Every request is a POST with a JSON body and:

text

Authorization: Bearer $SCAVIO_API_KEY

Endpoints

Base URL: https://api.scavio.dev. All paths are under /api/v1/g2. Every endpoint costs 5 credits.

Endpoint	Credits	What it returns
POST /api/v1/g2/search	5	Ranked software products, each row carrying product_id and slug
POST /api/v1/g2/product	5	The full profile - but no review text
POST /api/v1/g2/reviews	5	Reviews plus exact per-star counts and faceted counts

Workflow

1. Find the product: call /g2/search with query. Every row carries product_id and slug.
2. Profile: call /g2/product with product_id (a slug like notion, or the numeric G2 id like 82623 as a string - both resolve).
3. Reviews: call /g2/reviews with the same product_id.

/product carries no reviews at all. G2 loads review bodies in a separate frame, so the profile page simply does not contain them. If the user wants what customers said, you must call /reviews - there is no way to get it out of /product.

Conversely, /reviews carries things the profile has no form of: exact per-star counts, pros and cons with per-theme counts, and company-size / role / industry / region / category facets with counts. For a "how do enterprise buyers rate this" question, /reviews is the right call, not /product.

Every reference endpoint also accepts a full g2.com URL as url instead of an id.

Pagination

• /search - page (1-based) with limit (1-100, default 20). The 100 ceiling is ours, to keep a single request inside the request deadline; G2 itself keeps paginating at any size.
• /reviews - page, fixed at 10 per page, and it paginates well past the 10 pages G2's own widget links to.
• /product does not paginate.

At 5 credits a page, a 10-page review pull is 50 credits. Say so before starting.

Parameters

Search (/search)

Parameter	Type	Default	Description
query	string	one of	Search term (1-200 chars)
url	string	one of	Full g2.com/search URL (1-1000 chars); the host is checked
page	integer	--	1-based; page size follows limit
limit	integer	20	1-100
sort	string	relevance	relevance, popular, alphabetical, rating
rating	integer	--	1-5. Products at or above this star rating.

query or url is required.

Product (/product)

Parameter	Type	Default	Description
product_id	string	one of	Slug (notion) or numeric G2 id as a string (82623) - both resolve (1-200 chars)
url	string	one of	Full g2.com product URL

product_id or url is required.

Reviews (/reviews)

Parameter	Type	Default	Description
product_id	string	one of	Slug or numeric id as a string
url	string	one of	Full g2.com reviews URL
page	integer	--	10 reviews per page, fixed
sort	string	relevance	relevance, newest, most_helpful, rating_high, rating_low
rating	integer	--	1-5, half-star-inclusive: 1 returns 0, 0.5 and 1-star reviews
company_size	string	--	small_business (<=50), mid_market (51-1000), enterprise (>1000)
role	string	--	user, administrator, executive_sponsor, internal_consultant, consultant, agency, industry_analyst
region	string	--	north_america, europe, asia, latin_america, anz, middle_east, africa
query	string	--	Full-text search within the reviews. Narrows the list and every facet count.

product_id or url is required.

Why every filter is a closed enum

Two upstream behaviours make free-text filters dangerous here, and both fail silently:

• An unknown sort is silently accepted. order=zzznotasort answers 200 with a full result set in some unstated ordering. The sort never ran and nothing in the response says so.
• An unknown filter value matches nothing. A bogus company-segment value returns "Reviews (0)", which reads exactly like "this product has no enterprise reviews".

So the enums above are the complete set. Do not pass a value outside them, and do not report a zero-result filtered call as a finding without re-checking it unfiltered.

Examples

python

import os, requests

BASE = "https://api.scavio.dev"
HEADERS = {"Authorization": f"Bearer {os.environ['SCAVIO_API_KEY']}"}

# 1. Find products - 5 credits
found = requests.post(f"{BASE}/api/v1/g2/search", headers=HEADERS,
    json={"query": "project management", "limit": 100, "sort": "rating", "rating": 4}).json()

# 2. Full profile - 5 credits. Slug or numeric id, both as strings.
profile = requests.post(f"{BASE}/api/v1/g2/product", headers=HEADERS,
    json={"product_id": "notion"}).json()

# NOTE: profile contains NO review text. For that you must call /reviews.

# 3. What enterprise admins say - 5 credits, 10 reviews, plus every facet count
reviews = requests.post(f"{BASE}/api/v1/g2/reviews", headers=HEADERS,
    json={"product_id": "notion", "company_size": "enterprise",
          "role": "administrator", "sort": "newest", "page": 1}).json()

# 4. Search the review text itself - narrows the list AND the facet counts
mentions = requests.post(f"{BASE}/api/v1/g2/reviews", headers=HEADERS,
    json={"product_id": "notion", "query": "migration", "page": 1}).json()

Response shapes

Every response uses the envelope { data, response_time, credits_used, credits_remaining }.

• search - star rating, review count, vendor, categories, seller description and logo, with product_id and slug on every row. total_results is G2's Products-tab headline and caps at 10000; total_by_type splits the query across products, sellers, categories and discussions.
• product - rating and per-star histogram, review count, vendor, description and seller website, pricing editions with parsed amounts, feature groups, categories and breadcrumbs, supported languages, integrations, alternatives, head-to-head comparisons, media, community discussions, and G2's AI-derived pros and cons. No review text.
• reviews - rating, title, likes and dislikes, problems solved, reviewer job title, industry and company size, validated and incentivized flags - plus exact per-star counts, pros and cons with per-theme counts, and company-size / role / industry / region / category facets with counts.

Guardrails

• State the cost. Every call is 5 credits. Before a multi-page pull, tell the user the total you intend to spend.
• Never loop /product over search results. Search already returns rating, review count, vendor and categories for every row.
• Never say "the profile has no reviews so this product has none". /product structurally cannot carry review text - call /reviews before drawing any conclusion about customer sentiment.
• total_results caps at 10000. Do not present it as an exact count of matching products.
• rating on /reviews is half-star-inclusive: rating: 1 includes 0 and 0.5-star reviews. Do not describe it as "exactly one star".
• Zero results from a filtered /reviews call is ambiguous - it may mean the filter matched nothing rather than the segment having no opinion. Re-check unfiltered before reporting it.
• Never fabricate product names, ratings, review counts, pricing or review text. Only return API data.

Failure handling

• 400 means an invalid or missing parameter - e.g. neither query/product_id nor url, or a value outside a closed enum. Fix and retry.
• 401 means the API key is invalid or missing. Check SCAVIO_API_KEY.
• 404 means the product does not resolve. Re-check the slug or id via /search.
• 429 means rate or usage limit exceeded. Wait before retrying. See https://scavio.dev/docs/rate-limits.
• 502 means G2 served a bot wall or a hollow shell. This one is billed - the upstream charged full price for a page that could not be used, and it arrives as a real HTTP 200 upstream before we classify it. Retries are deliberately conservative for exactly this reason: back off for several seconds and retry at most once or twice, then report the failure instead of burning credits.
• 503 means upstream is temporarily unavailable - wait a few seconds and retry.
• If SCAVIO_API_KEY is not set, prompt the user to export it before continuing.

Python SDK

langchain-scavio has no G2 tool - use the Scavio SDK directly:

bash

pip install scavio

python

from scavio import ScavioClient

client = ScavioClient()  # reads SCAVIO_API_KEY

found = client.g2.search(query="project management", limit=100, sort="rating", rating=4)
profile = client.g2.product(product_id="notion")          # no review text here
reviews = client.g2.reviews(product_id="notion", company_size="enterprise",
                            role="administrator", sort="newest", page=1)

JavaScript / TypeScript:

bash

npm install scavio

js

import { Scavio } from "scavio";

const scavio = new Scavio(); // reads SCAVIO_API_KEY
const reviews = await scavio.g2.reviews({ product_id: "notion", company_size: "enterprise" });