API quickstart

The platform exposes two machine surfaces: a bearer-authenticated REST API at https://newdocs.hank.ai/api/v1 and a Model Context Protocol (MCP) server at https://newdocs.hank.ai/mcp. Both run the same publish pipeline, so anything you publish is secret- and PHI-scanned, link-checked, and routed through review exactly like a human edit. This page takes you from zero to your first call; the full API reference documents every endpoint, and Connect a publisher covers the product context.

Get a token

Every call carries a Hank-issued API key with the hank-docs scope. Issue keys from the Hank Codes console and send the key as a bearer credential on every request:

Every request

Authorization: Bearer <your-hank-docs-token>

A missing or malformed token returns 401 (unauthorized); a valid token without the hank-docs scope returns 403 (forbidden). A few operator-only routes additionally require operator privilege. Most integrations never touch them.

Your first call

GET https://newdocs.hank.ai/api/v1/topics returns a paginated list of topics. Page through it with ?limit and ?offset, and optionally filter by ?status= or ?sourceKey=.

cURL

curl https://newdocs.hank.ai/api/v1/topics?limit=20 \
  -H "Authorization: Bearer $HANK_DOCS_TOKEN"

Python (requests)

import os
import requests

token = os.environ["HANK_DOCS_TOKEN"]
resp = requests.get(
    "https://newdocs.hank.ai/api/v1/topics",
    headers={"Authorization": f"Bearer {token}"},
    params={"limit": 20, "offset": 0},
)
resp.raise_for_status()
page = resp.json()

JavaScript (fetch)

const token = process.env.HANK_DOCS_TOKEN;
const resp = await fetch("https://newdocs.hank.ai/api/v1/topics?limit=20", {
  headers: { Authorization: "Bearer " + token },
});
const page = await resp.json();

The response is a page envelope. Read hasMore (and advance offset by limit) to walk every page:

200 OK

{
  "items": [],
  "limit": 20,
  "offset": 0,
  "total": 128,
  "hasMore": true
}

Publish a topic

POST https://newdocs.hank.ai/api/v1/topics/upsert is the idempotent publish path. Key it on a stable sourceKey: re-running the same upstream document updates the same topic instead of forking a duplicate. A brand-new sourceKey returns 201 Created; an update to an existing one returns 200 OK.

cURL

curl -X POST https://newdocs.hank.ai/api/v1/topics/upsert \
  -H "Authorization: Bearer $HANK_DOCS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "manualId": "00000000-0000-0000-0000-000000000000",
    "slug": "rate-limits",
    "title": "Rate limits",
    "excerpt": "How request quotas work.",
    "bodyMdx": "# Rate limits: how request quotas work.",
    "sourceKey": "kb/rate-limits",
    "visibility": "public"
  }'

Python (requests)

import os
import requests

token = os.environ["HANK_DOCS_TOKEN"]
topic = {
    "manualId": "00000000-0000-0000-0000-000000000000",
    "slug": "rate-limits",
    "title": "Rate limits",
    "excerpt": "How request quotas work.",
    "bodyMdx": "# Rate limits: how request quotas work.",
    "sourceKey": "kb/rate-limits",
    "visibility": "public",
}
resp = requests.post(
    "https://newdocs.hank.ai/api/v1/topics/upsert",
    headers={"Authorization": f"Bearer {token}"},
    json=topic,
)
print(resp.status_code)  # 201 created, or 200 on an idempotent update

JavaScript (fetch)

const token = process.env.HANK_DOCS_TOKEN;
const resp = await fetch("https://newdocs.hank.ai/api/v1/topics/upsert", {
  method: "POST",
  headers: {
    Authorization: "Bearer " + token,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    manualId: "00000000-0000-0000-0000-000000000000",
    slug: "rate-limits",
    title: "Rate limits",
    excerpt: "How request quotas work.",
    bodyMdx: "# Rate limits: how request quotas work.",
    sourceKey: "kb/rate-limits",
    visibility: "public",
  }),
});

Handle errors

Every error response carries the same envelope: a human-readable error message and a stable machine code. Branch on the code, never on the message text, which can change.

Error envelope

{ "error": "manualId references a manual that does not exist", "code": "unknown_parent" }

The full code catalog:

`code`	HTTP	When you get it
`invalid_request`	400	The body, query, or path param failed validation.
`duplicate_slug`	400	A unique slug or `sourceKey` collides with an existing topic.
`unknown_parent`	400	A referenced product, manual, or chapter does not exist.
`content_rejected`	400	The content failed the publish pipeline. Fix the content, not the request shape.
`unauthorized`	401	Missing, malformed, or unrecognized bearer token.
`forbidden`	403	A valid token lacking the `hank-docs` scope or the required privilege.
`not_found`	404	The addressed resource does not exist.
`payload_too_large`	413	The request body exceeds the size cap.
`rate_limited`	429	The write rate limit was exceeded. See `Retry-After`.

Respect rate limits

Write responses carry X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset (epoch seconds). Exceed the quota and the API returns 429 (rate_limited) with a Retry-After header. Wait that many seconds, then retry. GET reads are not write-rate-limited.

Back off on 429

import time
import requests

def upsert_with_backoff(topic, token, attempts=5):
    delay = 1.0
    for _ in range(attempts):
        resp = requests.post(
            "https://newdocs.hank.ai/api/v1/topics/upsert",
            headers={"Authorization": f"Bearer {token}"},
            json=topic,
        )
        if resp.status_code != 429:
            return resp
        # Honor Retry-After when present; otherwise back off exponentially.
        wait = float(resp.headers.get("Retry-After", delay))
        time.sleep(wait)
        delay *= 2
    raise RuntimeError("still rate limited after retries")

Call the MCP server

Connecting an agent (Claude Code or another MCP client)? The Connect Claude Code guide has the claude mcp add line and the .mcp.json block, generated from the same helper this server advertises.

The MCP server at https://newdocs.hank.ai/mcp speaks JSON-RPC 2.0 over HTTP. The handshake is initialize → tools/list → tools/call.

Read tools: search_docs, ask_docs, list_docs, list_manuals, get_page, get_section, get_related, list_changelog, list_api_specs, get_openapi, list_spec_versions, diff_openapi, list_glossary, get_glossary, get_analytics, get_my_usage, get_topic_status, list_my_submissions, get_topic_preview, verify_audit, and list_audit. Org-scoped ticket tools (list_my_tickets, get_ticket, get_ticket_summary, and search_tickets) read only your own organization’s tickets.

Write tools: upsert_topic writes a topic through the same pipeline and force-pend as the REST upsert, and an OPERATOR token (is_admin) may pass requireReview: false to publish directly. publish_changelog, create_product, create_manual, create_chapter, and upload_image each require an operator token. There is no delete tool: destructive verbs stay off MCP.

initialize

curl -X POST https://newdocs.hank.ai/mcp \
  -H "Authorization: Bearer $HANK_DOCS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "jsonrpc": "2.0", "id": 1, "method": "initialize", "params": {} }'

tools/call: search_docs

curl -X POST https://newdocs.hank.ai/mcp \
  -H "Authorization: Bearer $HANK_DOCS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": { "name": "search_docs", "arguments": { "query": "rate limits" } }
  }'

Next steps

Browse every endpoint and schema in the API reference.
Manage the full resource lifecycle (create, update, publish, delete) in the resource lifecycle guide.
Building an autonomous agent? See the agent integration guide.
Generate a client from the OpenAPI 3.1 document.
Track releases in the changelog.
For product context, read Connect a publisher.

Loading the quickstart…