Describe the work.Watch agents run it.
The no-code AI agent builder. Design workflows on a canvas with any model and the apps you already use, watch every node run live, and replay any run.
Works with the models and tools you already use
Claude
OpenAI
Gemini
Mistral
Grok
GitHub
Notion
MCP
Agent glue code took weeks.Now it's one sentence.
Every team has that script. A cron job, three API keys and a prompt nobody dares to touch.
Label and route new issues
issue-triage
- New issue
- Classify
- Label
- Assign
Summarize #support every morning
daily-digest
- Schedule
- Read channel
- Summarize
- Post
Research three competitors
market-scan
- Topic
- Web search
- Agent
- Report
File notes into Notion
meeting-notes
- Transcript
- Extract
- Create page
Turn reports into Linear issues
bug-intake
- Webhook
- Dedupe
- Create issue
Draft replies to customer mail
inbox-zero
- New mail
- Agent
- Draft
- Review
Ask three models, keep the best
model-compare
- Prompt
- GPT
- Gemini
- Judge
One canvasfor every step.
Any model
Agents with tools
Structured output
Apps and MCP
Live runs
Replay
Describe it.Get a graph you can run.
When a GitHub issue opens, classify it, then post urgent ones to Slack
EnterDraft ready · 5 nodes · validated
Give your AI agentsaccess to Worfilo.
One install connects Worfilo to your editor. Your agent plans, builds, validates and runs workflows for you.
Claude Code
Cursor
VS Code
Windsurf
- $npx @worfilo/mcp install --client claude-code
- Added Worfilo to Claude Code
- > Build a workflow that triages new GitHub issues into Linear
- list_node_types 14 types
- create_workflow issue-triage (4 nodes)
- validate_workflow no issues
- run_workflow succeeded, Linear issue LIN-482 created
Every workflowis an endpoint.
Publish to freeze a version, so editing never changes what production runs. Then call it from your code.
- Post an input, read the output from the response
- Wait for the result, poll the run, or stream every node over SSE
- Scope a key to chosen workflows, shown once and stored as a hash
Request
curlcurl https://api.worfilo.com/v1/workflows/WORKFLOW_ID/runs \
-H "Authorization: Bearer $WORFILO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"input": {
"issue": "Checkout fails on Safari"
}
}'Response
SSE stream- node.succeededtrigger
- node.succeededclassify
- node.succeededis_urgent
- run.succeeded2.4s · 1,183 tokens
Handling secrets?Private by default.
Model keys, app tokens and API secrets are encrypted, never placed in a workflow, and never returned to the browser. Private network addresses are refused. And the PII Shield masks personal data before an agent ever reads it.
- ANTHROPIC_API_KEY
- GITHUB_TOKEN
- SLACK_BOT_TOKEN
Run record
headers: {
authorization: "••••••"
}
status: 200Ticket text
Hi, I'm Nimal Perera. My card 4111 1111 1111 1111 was charged twice. Call +94 77 123 4567.
What the agent reads
Hi, I'm <PERSON_1>. My card <CREDIT_CARD_1> was charged twice. Call <PHONE_1>.
Encrypted at rest
Model keys, app tokens and API secrets are stored encrypted.
Write-only
Secrets are never returned to the browser or placed in a workflow.
Scrubbed from runs
Run records keep results and status, never credentials.
PII masked first
Names, emails, phone numbers, card numbers and IDs become tokens before the model reads them.
- Masking profiles
- General
- Healthcare
- Finance
- Minimal
Questions,answered.
Describe it.Watch it run.
Your first workflow is one sentence away. Pick a model, connect an app, and ship it as an API when it works.