A snippet is copied from docs.blitz-api.ai and carries the URL it came from. The surface of an SDK is not deduced by analogy.
Blitz Brand Lab · Code
Code on the site is quoted, never invented
Six surfaces show code, from the developers hub to a lesson payload. Every one of them follows the same contract: the real base URL, the real header, field names copied from the API reference, and a source URL for every snippet.
15 PUBLISHED ENDPOINTSBASE URL api.blitz-api.aiHEADER x-api-keyGEIST MONO · INK PANEL
A path written « by convention » before its doc exists is a bug on every surface: an invented enrichment path lived on the product page, the use case frame and llms-full.txt for 19 days.
person_linkedin_url, people.min_per_company, max_results, cursor, fair_usage: the names are the API’s, not the copy’s.
Geist Mono, 11 to 13px, on the ink panel (--color-ink, --radius-l). Labels uppercase, .06 to .08em. No syntax highlighting theme, no line wrapping.
01
6 SURFACESWhere code appears
Six surfaces, one look
Each surface has its fiche on the Sections, Content or Assets page. The table says what it shows and where it lives.
| Surface | File | Where | What it shows | Fiche |
|---|---|---|---|---|
| CodeTabs | hub/CodeTabs.astro | Developers hub | The same call in cURL, Python and TypeScript. Accessible tabs, no-JS safe, one verbatim source URL per snippet. | Open → |
| CodeSplit panel | product/CodeSplit.astro | Product opening A | The dark 556×448 panel receives an ASSET, not text: the code idea is drawn (api-request, terminal-command), never pasted. | Open → |
| Payload | learn/blocks/Payload.astro | Lessons | The annotated code block: numbered lines, notes anchored on the lines to adapt. | Open → |
| Endpoint | content/Endpoint.astro | Articles and lessons | The card of an endpoint: method, path, the parameters that matter, the response fields to read, the reference link. | Open → |
| api-request · terminal-command · json-* | visual-assets/BlitzAsset.astro | Any proof frame | Code as a sign: a method chip and a path, a prompt and a command, a typed key-value pair. | Open → |
| LabCode | layout-system/LabCode.astro | This lab | The reference block of the lab: label bar, copy button, ink panel. Not shipped on public pages. | Open → |
02
3 BLOCKSThe panel and the blocks
Ink panel, annotated when it teaches
A plain block states; a Payload teaches (three notes for fifteen lines); an Endpoint card lists only the fields the text talks about. The three below use the official Person Enrichment and TAM By People references.
POST https://api.blitz-api.ai/v2/enrichment/person
x-api-key: YOUR_API_KEY
Content-Type: application/json
{ "person_linkedin_url": "https://www.linkedin.com/in/antoine-blitz-5581b7373" }
{
"people": {
"job_title": { "include": ["Sales"] },
1
"location": { "country_code": ["US"] },
"min_per_company": 3
2
},
"company": {
"industry": { "include": ["Software Development"] },
"employee_range": ["51-200", "201-500"]
},
"max_results": 50,
3
"cursor": null
}
- Keyword match by default; wrap a title in [brackets] for an exact match.
- The persona floor: 1 to 25, 0 = unset. A heavy floor can return partial pages: keep paging.
- Up to 50 companies per page, 10 by default. One result per company returned.
Retrieve a professional’s full profile and entire career history from their LinkedIn profile URL.
Send
person_linkedin_urlrequired- LinkedIn profile URL of the person to enrich
Read
found- true when the profile resolved; false returns person: null and counts nothing
person.experiences[]- Every role: company_linkedin_url, company_linkedin_id, company_domain, dates, job_is_current
fair_usage.request_id- Quote it when you contact support
03
7 FACTSAPI contract
What every snippet assumes
Read from the official OpenAPI v2 (docs.blitz-api.ai). If one of these facts changes, every surface above changes with it.
- Base URL
- https://api.blitz-api.ai
- Authentication
- x-api-key header, one key per workspace (app.blitz-api.ai)
- Method
- Every v2 endpoint is a POST with a JSON body (key info is the one GET)
- Pagination
- cursor-based: pass the returned cursor back until it comes back null; max_results up to 50
- Usage block
- fair_usage on every response: records_used, records_remaining, next_reset_at, rate_limit, request_id
- Not found
- found: false with a null object, nothing counted
- Errors
- 401 invalid key · 402 fair use limit reached · 404 not found · 429 too many requests
cURL · verbatim from the docs
curl -X POST https://api.blitz-api.ai/v2/enrichment/email \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"person_linkedin_url":"https://www.linkedin.com/in/example-person"}'Python · verbatim from the docs
from blitz_api import BlitzAPI
client = BlitzAPI() # reads BLITZ_API_KEY
email = client.enrichment.email(
person_linkedin_url="https://www.linkedin.com/in/example-person"
)
if email.found:
print(email.email)TypeScript · verbatim from the docs
import { BlitzAPI } from "blitz-api-js";
const client = new BlitzAPI(); // reads BLITZ_API_KEY
const email = await client.enrichment.email({
person_linkedin_url: "https://www.linkedin.com/in/example-person",
});
if (email.found) console.log(email.email);04
15 ENDPOINTSPublished endpoints
The catalog the site speaks
One line per product, from the endpoint field of the product JSON: the same field that renders the Endpoint line of /llms-full.txt and the subtitle of the steps sections.
Machine surfaces: /llms.txt (curated index) and /llms-full.txt (the whole site in one markdown file), both generated at build from the same sources; the API reference itself lives at docs.blitz-api.ai.
05
4 GRAMMARSSite code conventions
Data-cta ids, configuration, schemas
The code of the site itself follows a few closed grammars. They live in config files and are checked at build or by the audit.
data-cta ids
Every conversion link carries data-cta="scope.location.intent[.plan]", built by ctaId() in config/tracking.ts. The lists are closed (tracking-contract.json); an id outside them fails the build, and the audit counts CTAs without an id.
TS
ctaId('product', 'hero', 'primary') // → "product.hero.primary"
ctaId('pricing', 'pricing_card', 'choose', 'full') // → "pricing.pricing_card.choose.full"
scopes: global · home · pricing · product · use_case · team · developers · integrations · about · contact · legal · other
locations: nav · nav_mobile · hero · pricing_card · plan_matrix · plan_compare · embedded_band · cta_section · inline · footer · unknown
plans: core · plus · fullContent schemas
src/content.config.ts types every collection: products (family, endpoint, seo, hero, sections, faq, docsUrl, relatedProducts), useCases, teams, blog, courses, lessons. A section kind outside the enum fails the build; a missing optional field does not, which is why the templates list what each section requires.
TS · content.config.ts
const section = z.object({
kind: z.enum(['features', 'showcase', 'split', 'steps', 'usecases', 'apis']),
title: z.string().optional(),
subtitle: z.string().optional(),
asset: z.string().optional(), // BlitzAssetId
signature: z.string().optional(), // BlitzSignatureId
items: z.array(sectionItem),
});Configuration files
config/site.ts: NAV, PRODUCT_FAMILIES, USE_CASES, TEAMS, FOOTER, EXTERNAL, CONTACT_FORM, PRIVACY_FORMconfig/pricing.ts: SEPT, PLANS, PRICING_FAQ, EMBEDDED, COPY: every offer figureconfig/signature-bands.ts: FAMILY_BANDS, USECASE_BANDS, HUB_BANDS, TEAM_BANDSconfig/tracking.ts+tracking-contract.json: loader, consent, data-cta grammarconfig/testimonials.ts: slides, QUOTES_ARE_PLACEHOLDERS guard
Build and checks
npm run build: 47 pages, sitemap, llms.txt and llms-full.txtpython3 scripts/audit-sept.py dist: links, banned routes, forbidden terms, SEO bounds, JSON-LD, offer facts, sitemap, trackingnode scripts/rhythm-harness.mjs: rails, gutter, vertical intervals, overflow at 1440 / 991 / 767 / 375node scripts/capture.mjs <url> <width> <out.png>: full-page capture with real device emulation- Push on main deploys to production (Vercel) in about two minutes: build and audit first, always