marktoflow
Health Pass
- License — License: AGPL-3.0
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 19 GitHub stars
Code Fail
- network request — Outbound network request in examples/custom-integration/integrations/weather-api.ts
- rm -rf — Recursive force deletion command in package.json
- network request — Outbound network request in package.json
Permissions Pass
- Permissions — No dangerous permissions requested
This tool is an AI-powered workflow automation framework that lets users define and execute automation sequences using simple markdown and YAML files, leveraging existing AI subscriptions like Copilot or Claude.
Security Assessment
Risk Rating: Medium. The tool requires careful review before implementation. An automated scan failed due to a recursive force deletion command (`rm -rf`) found in the `package.json` file. While this is often used for cleaning build directories, it is a potentially destructive operation if an error occurs in the script path. Additionally, the codebase makes outbound network requests from both the core package and an example weather integration. Users should verify where these requests are going to ensure no sensitive data is leaked. No hardcoded secrets were found, and the tool does not request explicitly dangerous system permissions.
Quality Assessment
The project is relatively new and has a low community trust level, currently sitting at 19 GitHub stars. However, it is actively maintained, with the most recent push occurring today. It is legally safe to use, review, and contribute to, as it is properly licensed under AGPL-3.0.
Verdict
Use with caution — investigate the `rm -rf` script and network activity in the source code before deploying.
Automation frameworks shouldn't fight you. Marktoflow is CLI-native automation with the power of AI agents. Write workflows in markdown+YAML, execute with Claude Code, GitHub Copilot, Codex, or OpenCode — no AI API keys required. 39 direct SDK integrations, visual designer UI, and full MCP compatibility. Simple to write, powerful to run.
marktoflow
Open-source AI workflow automation — your workflows are just markdown files.
Use your Copilot/Claude/Codex subscriptions · 39 integrations · Tool calling · Local LLMs · Self-hosted
Visual workflow designer — drag-and-drop nodes, AI-assisted editing, one-click execution
Quick Start
npm install -g marktoflow
marktoflow init
marktoflow run workflow.md
That's it. Your workflow is a markdown file with YAML frontmatter:
---
workflow:
id: hello-world
name: Hello World
tools:
slack:
sdk: '@slack/web-api'
auth:
token: '${SLACK_BOT_TOKEN}'
steps:
- action: slack.chat.postMessage
inputs:
channel: '#general'
text: 'Hello from marktoflow!'
---
Already have a GitHub Copilot, Claude, or Codex subscription? Use it directly — no API keys:
marktoflow run workflow.md --agent copilot # GitHub Copilot subscription
marktoflow run workflow.md --agent claude # Claude CLI login
marktoflow run workflow.md --agent codex # Codex CLI
marktoflow run workflow.md --agent ollama # Free, runs locally
marktoflow run workflow.md --agent gemini # Google Gemini subscription
Why marktoflow?
| What you get | |
|---|---|
| Markdown-native | Workflows are .md files — readable, auditable, diffable, version-controlled |
| Use your existing AI | Copilot, Claude, Codex subscriptions — no extra API keys needed |
| AI tool calling | Agentic loops — models call tools, get results, and reason autonomously |
| Local LLMs too | Optional llama.cpp, VLLM, Ollama support for air-gapped or offline use |
| Secure by design | Self-hosted, no telemetry, your data never leaves your infrastructure |
| Structured output | JSON mode and JSON Schema validation for reliable AI responses |
| MCP-first | Native Model Context Protocol support with zero config |
| Direct SDK calls | Official SDKs, full TypeScript types, no wrapper APIs |
| Parallel execution | Run multiple AI agents concurrently — 3x faster code reviews, batch processing |
| Visual editor | Optional drag-and-drop GUI with marktoflow gui |
| Cost tracking | Know exactly what each workflow run costs |
Security & Privacy
marktoflow is designed for teams and individuals who care about where their data goes.
| marktoflow | Cloud agent platforms | |
|---|---|---|
| Data residency | Your infrastructure — nothing leaves unless you send it | Data processed on vendor servers |
| Local LLM support | First-class — llama.cpp, VLLM, Ollama | Typically cloud-only |
| Air-gapped operation | ✅ Full offline support with local models | ❌ Requires internet |
| Audit trail | Git history + SQLite execution logs — fully inspectable | Vendor-controlled logs |
| Workflow transparency | Plain markdown — read every step before it runs | Opaque agent behavior |
| Telemetry | None — zero phone-home, zero tracking | Varies |
| Credential storage | Local encryption (AES-256-GCM) or your vault (HashiCorp, AWS, Azure) | Vendor-managed |
| Source available | AGPL-3.0 — inspect, modify, self-host | Often proprietary |
Use case: You want AI-powered automation but your company policy (or common sense) says sensitive data — code, credentials, customer info, internal docs — shouldn't flow through third-party agent platforms. marktoflow runs entirely on your machine or your servers with your choice of LLM.
How it compares
vs. Workflow Platforms
| Feature | marktoflow | Zapier | n8n | GitHub Actions |
|---|---|---|---|---|
| Open source | Yes | No | Yes | No |
| Workflow format | Markdown | Proprietary | JSON | YAML |
| Version control | Git-native | No | Limited | Git-native |
| AI agent support | Built-in | Add-on | Plugin | Limited |
| AI tool calling | Built-in | No | Plugin | No |
| Local LLMs | Yes | No | Via plugin | No |
| Direct SDK access | Yes | No | No | Via actions |
| Visual editor | Yes | Yes | Yes | No |
| Self-hosted | Yes | No | Yes | Runners only |
| Per-task pricing | Free | Yes | Free (self-host) | Minutes-based |
vs. AI Agent Platforms
| Feature | marktoflow | OpenClaw | LangChain | CrewAI |
|---|---|---|---|---|
| Self-hosted | Yes | Gateway only | Yes | Yes |
| Local LLM support | First-class | Limited | Yes | Yes |
| Air-gapped / offline | Yes | No | Partial | Partial |
| No telemetry | Yes | No | Opt-out | Opt-out |
| Workflow format | Markdown (auditable) | Opaque | Python code | Python code |
| Visual editor | Yes | No | No | No |
| 39 native integrations | Yes | No | Community | Community |
| Non-AI workflows | Yes | No | No | No |
| Data stays local | Always | Routed through gateway | Depends on LLM | Depends on LLM |
Integrations
39 native SDK integrations — all with TypeScript types, retry logic, and input validation.
| Category | Services |
|---|---|
| Communication | Slack, Teams, Discord, Telegram, WhatsApp, Twilio |
| Gmail, Outlook, SendGrid, Mailchimp | |
| Google Workspace | Sheets, Calendar, Drive, Docs |
| Project Management | Jira, Linear, Asana, Trello |
| Knowledge | Notion, Confluence |
| Developer | GitHub, Airtable |
| Payments | Stripe, Shopify |
| Support | Zendesk |
| Storage | Dropbox, AWS S3 |
| Databases | Supabase, PostgreSQL, MySQL |
| Universal | HTTP client (any REST API), RSS/Atom feeds |
| AI Agents | GitHub Copilot, Claude, Codex, Gemini — use your existing subscriptions, no API keys — plus OpenAI, Ollama, llama.cpp, VLLM for local/custom setups — control via Slack/Telegram |
Packages
| Package | Description |
|---|---|
marktoflow |
All-in-one install: npm install -g marktoflow |
@marktoflow/marktoflow |
Scoped metapackage (CLI + GUI + integrations) |
@marktoflow/core |
Parser, engine, state management, plugin system |
@marktoflow/cli |
Command-line interface and workflow runner |
@marktoflow/gui |
Visual workflow designer (web UI) |
@marktoflow/integrations |
39 service integrations and AI adapters |
Examples
Production-ready workflow templates in examples/:
agent-task-executor — Control AI agents (Claude, Copilot, OpenCode, Ollama, Gemini) via Slack/Telegram. Send task instructions, get structured results with pass/fail status.
Example: Send "Create a React component called UserProfile with email validation" → Agent executes with safe permissions → Returns structured results (3/3 tasks passed ✅)
# Run the Slack version marktoflow run examples/agent-task-executor/workflow-slack.md # Or use Telegram marktoflow run examples/agent-task-executor/workflow-telegram.mdparallel-agents — ⚡ NEW: Run multiple AI agents concurrently
- multi-agent-code-review — Security + Performance + Quality reviews in parallel (3x faster)
- batch-pr-processing — Review 50 PRs in 5 minutes instead of 25
- consensus-decision — Gather diverse AI perspectives for better decisions
codebase-qa — AI-powered Q&A via Slack/Telegram
copilot-code-review — PR review with GitHub Copilot
daily-standup — Jira + Slack standup automation
incident-response — Multi-service incident coordination
approval-workflow — Human-in-the-loop with web forms
sprint-planning — AI-assisted sprint planning
Documentation
📋 CHANGELOG — Version history, new features, security fixes, breaking changes
- Installation Guide
- Detailed Guide
- YAML API Reference — Complete API including ⚡ parallel execution
- REST API Guide
- Template Expressions
- Control Flow
- GUI User Guide
Community
- GitHub Discussions — Questions, ideas, show & tell
- Contributing — PRs welcome
- Issues — Bug reports and feature requests
License
AGPL-3.0 — Free for personal and open source use. Commercial licensing available for organizations that cannot comply with AGPL terms. Contact [email protected] for details.
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found