abstractflow 0.3.1

Diagram-based AI workflow generation built on AbstractCore

AbstractFlow

Diagram-based, durable AI workflows for Python.

AbstractFlow provides:

• A portable workflow format (VisualFlow JSON) and helpers to execute it from any host (abstractflow.visual).
• A simple programmatic API (Flow, FlowRunner) backed by AbstractRuntime.
• A reference visual editor app in web/ (FastAPI backend + React frontend).

Project status: Pre-alpha (pyproject.toml: Development Status :: 2 - Pre-Alpha). Expect breaking changes.

Capabilities (implemented)

• Execute programmatic flows (Flow → FlowRunner) with a default in-memory runtime.
• Execute portable VisualFlow JSON from any host process (abstractflow.visual).
• Durable waits and resumption via AbstractRuntime (e.g. user/event/schedule waits).
• Package a flow tree as a WorkflowBundle (.flow) via the CLI.
• Author/run VisualFlows in the reference web editor (web/).

Evidence (code): abstractflow/runner.py, abstractflow/visual/executor.py, abstractflow/cli.py, web/backend/routes/ws.py.

Docs

• Start here: docs/getting-started.md
• API reference: docs/api.md
• VisualFlow format: docs/visualflow.md
• Visual editor: docs/web-editor.md
• CLI: docs/cli.md
• FAQ: docs/faq.md
• Architecture: docs/architecture.md
• Docs index: docs/README.md

Installation

pip install abstractflow

Requirements: Python 3.10+ (pyproject.toml: requires-python).

Optional extras:

• Agent nodes (ReAct workflows): pip install "abstractflow[agent]"
• Dev tools (tests/formatting): pip install "abstractflow[dev]"

Notes:

• abstractflow depends on AbstractRuntime and abstractcore[tools] (see pyproject.toml).
• Some VisualFlow node types require additional packages (e.g. memory_kg_* nodes need abstractmemory).

Quickstart (programmatic)

from abstractflow import Flow, FlowRunner

flow = Flow("linear")
flow.add_node("double", lambda x: x * 2, input_key="value", output_key="doubled")
flow.add_node("add_ten", lambda x: x + 10, input_key="doubled", output_key="final")
flow.add_edge("double", "add_ten")
flow.set_entry("double")

result = FlowRunner(flow).run({"value": 5})
print(result)  # {"success": True, "result": 20}

Quickstart (execute a VisualFlow JSON)

import json
from abstractflow.visual import VisualFlow, execute_visual_flow

with open("my-flow.json", "r", encoding="utf-8") as f:
    vf = VisualFlow.model_validate(json.load(f))
result = execute_visual_flow(vf, {"prompt": "Hello"}, flows={vf.id: vf})
print(result)  # {"success": True, "waiting": False, "result": ...}

If your flow uses subflows, load all referenced *.json into the flows={...} mapping (see docs/getting-started.md).

Visual editor (from source)

The visual editor is a dev/reference app in web/ (not shipped as a Python package on PyPI).

git clone https://github.com/lpalbou/AbstractFlow.git
cd AbstractFlow

python -m venv .venv
source .venv/bin/activate
pip install -e ".[server,agent]"

# Terminal 1: Backend (FastAPI)
cd web && python -m backend --reload --port 8080

# Terminal 2: Frontend (Vite)
cd web/frontend && npm install && npm run dev

Open the frontend at http://localhost:3003 (default Vite port). See docs/web-editor.md.

CLI (WorkflowBundle .flow)

abstractflow bundle pack web/flows/ac-echo.json --out /tmp/ac-echo.flow
abstractflow bundle inspect /tmp/ac-echo.flow
abstractflow bundle unpack /tmp/ac-echo.flow --dir /tmp/ac-echo

See docs/cli.md and abstractflow/cli.py.

Related projects

• AbstractRuntime (durable execution kernel): https://github.com/lpalbou/AbstractRuntime
• AbstractCore (providers/models/tools): https://github.com/lpalbou/AbstractCore
• AbstractAgent (ReAct/CodeAct): https://github.com/lpalbou/AbstractAgent

Changelog

See CHANGELOG.md.

Contributing

See CONTRIBUTING.md.

Security

See SECURITY.md.

Acknowledgments

See ACKNOWLEDMENTS.md.

License

MIT. See LICENSE.