§01  /  Documentation

Everything you need
to build with Lyhnis.

Quickstarts, guides, REST and SDK reference, and integration playbooks. This is the shape of the documentation - sections marked soon are being written now.

/ ⌘K
70+ activitiesREST v1 · stableSDK v2.37Status: all systems nominal
In progress

The catalog and reference are being written activity by activity. Until a section lands, the team will answer directly - usually the same day.

Ask us instead
§02  /  Quickstarts

Pick a starting point.

Three paths in, each one to a running workflow inside ten minutes. The canvas is fastest, the SDK is most expressive, MCP is for connecting an agent you already have.

~5 minVisual canvas

Author your first workflow visually.

Drag activities onto the canvas, wire them up, hit Run. No code, no install - the fastest path to a running workflow.

1Sign in to app.lyhnis.com
2New workflow → drag activities
3Hit Run → inspect the run
Start the visual quickstartSoon
~10 minPython SDK

Author your first workflow in Python.

Install the SDK, decorate a function, run it locally. Workflows version with your code, diff like your code, deploy like your code.

1pip install lyhnis
2@workflow("my_first")
3lyhnis deploy & run
Start the SDK quickstartSoon
~15 minMCP connect

Connect an external agent over MCP.

Expose your workflows as tools to Claude, ChatGPT or any MCP-compatible client. Per-tenant server, JWT-bridge auth, audit by construction.

1Provision your tenant MCP URL
2Generate the JWT-bridge token
3Add to client → tools appear
Start the MCP quickstartSoon
§03  /  Browse by topic

Documentation
by surface.

Eight top-level sections, shaped like the product itself - what you see in the dashboard maps to what you find here.

Authoring

9 articles
  • Visual canvas basicsSoon
  • Workflow definition (JSON)Soon
  • Python SDK quickstartSoon
  • Custom activitiesSoon
  • Versioning & deploysSoon
Browse all in AuthoringSoon

Activities

11 categories · 70+
  • AI / LLM activitiesSoon
  • OCR activitiesSoon
  • HTTP & webhooksSoon
  • SaaS integrationsSoon
  • Data transformsSoon
Full catalogSoon

Agents

7 articles
  • agent.run referenceSoon
  • Four engine-enforced budgetsSoon
  • MCP inbound (expose workflows)Soon
  • MCP outbound (call external)Soon
  • Sub-workflows as toolsSoon
Browse all in AgentsSoon

Operations

8 articles
  • Dashboard & monitoringSoon
  • Audit logSoon
  • Communications trackingSoon
  • Files & lineageSoon
  • Prometheus metricsSoon
Browse all in OperationsSoon

Security & identity

6 articles
  • MFA setup & recovery codesSoon
  • SSO (SAML, OIDC, SCIM)Soon
  • Step-up auth on destructive actionsSoon
  • Workspace policiesSoon
  • Encrypted credential vaultSoon
Browse all in SecuritySoon

Integrations

28 SaaS connectors
  • Slack & TeamsSoon
  • Stripe (8 operations)Soon
  • HubSpot (10 operations)Soon
  • Google WorkspaceSoon
  • Microsoft 365Soon
Full integrationsSoon

Reference

REST · SDK · CLI · Webhooks
  • REST API referenceSoon
  • Python SDK referenceSoon
  • CLI referenceSoon
  • Webhook events (signed)Soon
  • Errors & status codesSoon
Full referenceSoon

Deployment

5 articles
  • Multi-tenant cloud (US / EU)Soon
  • Single-tenant dedicatedSoon
  • On-prem deploymentSoon
  • Data residencySoon
  • Disaster recoverySoon
Browse all in DeploymentSoon
§04  /  Same operation, two languages

API and SDK
quick reference.

Everything available in the dashboard is available over REST and through the Python SDK. Here is list workflows in both - pick the surface that fits your stack.

REST · curlv1 · stable
curl https://api.lyhnis.com/v1/workflows \
  -H "Authorization: Bearer $LYHNIS_API_KEY" \
  -G -d limit=50

# → cursor-paginated
{
  "data": [
    { "id": "wf_01JX7",
      "name": "invoice_processing",
      "version": 14,
      "schedule": "0 9 * * MON" }
  ],
  "cursor": "eyJvZmZzZXQiOjUwfQ"
}
15 endpointscursor-paginatedsigned webhooks
Python SDKv2.37 · type-checked
from lyhnis import Client

client = Client(api_key=os.environ["LYHNIS_API_KEY"])

# iterates every page for you
for wf in client.workflows.list():
    print(wf.name, wf.version)

# or take the cursor yourself
page = client.workflows.list(limit=50)
while page.cursor:
    page = client.workflows.list(cursor=page.cursor)
mirrors REST 1:1async variantpydantic models
§05  /  Stay current

What's new,
what's working.

The latest releases, live platform status, and where to ask when the docs come up short.

Latest releasesweekly
v2.37 · Outbound MCP credentials26 May
v2.36 · Sub-workflows as tools12 May
v2.35 · Recursive cost rollup28 Apr
v2.34 · Communications14 Apr
Platform statusnominal
API200 ok · 61ms p95
Workers28 / 28 healthy
Queue lag0s
Uptime 90d99.98%
When docs fall shorthumans
support@lyhnis.comactive workspaces
sales@lyhnis.compricing · demos
security@lyhnis.comsecurity docs · disclosure
hello@lyhnis.comanything else