Guides

Step-by-step walkthroughs for building real workflows. Each guide ends with a deployed, working automation. Every guide is Markdown-first (agents can fetch raw MDX via /api/docs-markdown?path=... or the Copy for agent button) and includes a TSL block you can paste into Code mode or apply_tsl.

Recommended learning path (end-to-end)

Follow this order if you are new. A random logged-in user should be able to finish a full workflow using only these pages.

| Step | Guide | Outcome | |------|-------|--------| | 0 | Quick Start | Account, workspace, first canvas | | 1 | Send Email on Trigger | Manual trigger → email (proves env + secrets) | | 2 | Webhook Automation | External event → transform → action | | 3 | Build a REST API | Sync HTTP API + Return | | 4 | Choosing a Trigger | When to use Webhook vs dedicated vs Manual vs API | | 5 | Platform guides below | Slack / Telegram / GitHub / Google as needed | | 6 | Build an AI Agent | Agent path |

One-click starters: Quick-Start Templates (Echo, Hello World API, Proxy, etc.).

1. Getting started

Send Email on Trigger — Simplest useful workflow. Manual run sends a personalized email. Use this to verify secrets and Managed execution.

  • Uses: Manual Trigger, Code, Send Email (Resend)
  • Time: ~10 minutes

Quick-Start Templates — Seven ready graphs you can create in one click.

  • Uses: Webhook, Transform, HTTP Request, If, Code, Return, Stop
  • Time: ~5 minutes each

2. Automation (inbound events)

Webhook Automation — GitHub raw webhook (HMAC) → Transform → email. Long-tail / signature-verified HTTP only.

  • Uses: Webhook Trigger (provider=github), Transform, Code, Send Email
  • Time: ~15 minutes
  • Note: For GitHub App installs and installation-scoped events, use GitHub Trigger, not this path.

Telegram Bot Workflow — BotFather token → Telegram Trigger → reply.

Automation on Git Commits — GitHub-oriented automation patterns.

3. APIs

Build a REST API Endpoint — Authenticated HTTP endpoint, external call, synchronous JSON via Return.

  • Uses: API Trigger, HTTP Request, Transform, Return
  • Time: ~10 minutes

Build a CRUD API — Full GET/POST/PUT/DELETE style API on the canvas.

4. Choosing triggers (policy)

Choosing a Trigger — Decision table aligned with product policy:

| Need | Use | |------|-----| | OAuth / App install / watches (Google, Slack, GitHub App, Discord, Telegram) | Dedicated trigger | | Signed or unsigned HTTP long-tail (Stripe, Clerk, raw GitHub HMAC) | Webhook Trigger | | Canvas / CLI / subworkflow start | Manual Trigger | | Sync programmable REST | API Trigger | | Chat Completions compatible | Chat Trigger | | Expose workflow as MCP tool | MCP Trigger |

5. Compose & chain

Chain Workflows with Subworkflow — Parent + children; children usually start with Manual Trigger.

6. AI agents

Build an AI Agent — Tool calling, memory, OpenAI-compatible deploy.

Agent Memory — Window vs Qdrant.

Self-Host an AI Agent — CLI runner, local LLMs.

Deploy as an OpenAI Endpoint — OpenAI SDK compatible.

Process Local Files with an AI Agent — File tools on your machine.

Build a RAG System — Qdrant + agent.

Streaming AI Responses — SSE token streaming.

Build a Coding Agent — File Ops + Shell + Web Search.

7. Messaging & chatbots

Build a Messaging Chatbot — Telegram / Slack / Discord triggers + agent + send.

8. MCP

Expose a Workflow as an MCP Tool — MCP Trigger.

Connect MCP Servers — Attach external MCP to agents (prefer MCP Registry).

Tensorify MCP Server for AI IDEs — Cursor / Claude Desktop control plane (apply_tsl, deploy, test).

9. Local machine & DevOps

Local Machine Automation — File Ops, Shell on CLI runner.

Agent / Markdown notes

  • Copy for agent (top of every docs page) copies this page's raw content.mdx.
  • API: GET https://tensorify.io/api/docs-markdown?path=/docs/guides/...
  • Every guide should include a section ## TSL (Code mode / agents) with a complete, pasteable workflow.

Start with the Quick Start if you do not have a workspace yet.

On this page