Agenten-Kompass

AGENT PROFILES

Dify

LangGenius · Agent builders

Platform for building AI applications, agents and workflows with your own data. Applications can be published as web apps or through APIs.

Automate Research Organize office work

Dify · AI application and agent platform

Product and access

Dify provides the builder and runtime for custom AI applications. Agent nodes connect model-driven tool use to further workflow steps.

Provider

LangGenius, Inc. operates Dify Cloud and develops the Dify platform for AI applications. The Community Edition can be self-hosted; its license adds conditions for multi-tenant operation and visible branding to Apache 2.0.

Operator
LangGenius, Inc.
Organization
Company
Website
dify.ai
GitHub stars
157,962 · Reviewed
License
Dify Open Source License (Apache-2.0 with additional conditions)
Provider sourcesSource 1 Source 2 Source 3 Reviewed
01

Core features

8 features from provider sources · checked on 9 October 2026. Availability and limits are listed below.

  • Run agents within workflows

    An Agent node processes a task and passes results to subsequent steps.

    Source 2

  • Retrieve your own knowledge

    Knowledge bases supply context from processed documents to AI applications.

    Source 3

  • Configure model providers

    Supported model providers and their credentials can be configured in the workspace.

    Source 5

  • Publish as a web app or API

    Publishing makes applications available through interfaces or API access.

    Source 4

  • Inspect runs in logs

    Logs show application interactions and help inspect executions.

    Source 6

  • Visual application flows

    Workflow and Chatflow combine processing, knowledge and tools into AI applications.

    Source 1

  • Plugins and templates

    Custom plugins add models and tools; the marketplace provides integrations and app templates.

    Source 1

  • Applications as MCP servers

    Dify apps can be published as tools for external MCP clients.

    Source 10
02

Requirements and limits

  • Requirement

    Configure compatible models and tools

    The agent strategy needs a suitable model. External tools require credentials, descriptions and appropriate parameters.

    Source 2

  • Limitation

    Account for the new Agent node being in beta

    The new Agent node is documented as beta for Workflow apps. Text variables are truncated to 2,000 characters when passed in; longer content can be passed as a file.

    Source 2

  • Requirement

    Check the Dify license for your deployment

    The license is based on Apache 2.0 with additional conditions. Multi-tenant operation requires written authorization; frontend logos and copyright notices have specific requirements.

    Source 7

  • Requirement

    Enable MCP access explicitly

    The MCP server is disabled initially. Its generated URL contains credentials and must be treated like an API key.

    Source 10

Operation & oversight

Operational details for this profile have not yet been researched.

03

Practical tasks

Editorial examples based on documented features. No product tests performed.

01

Prepare a support reply with knowledge

Your input
Approved product documentation, customer question and desired output format.
Possible result
A reply draft from a workflow with knowledge retrieval.
What to check
Check references, answer completeness and handoff to the next step.
02

Turn documents into structured results

Your input
Document files, extraction task and defined output fields.
Possible result
An agent task with results for downstream processing.
What to check
Review file completeness, field values, execution logs and costs.
EDITORIAL EXAMPLE

Classify a support question using approved documents and produce a reply draft as workflow output.

Your possible prompt

Design a process for new requests: capture information, sort it by topic, and create a proposed response. Plan an approval step before external actions.

Getting-started idea based on documented capabilities. Not an executed product test. Adapt the task, data, and permissions.

04

Setup and pricing

Setup

Set up a workspace, connect model providers, and configure workflow, knowledge and tools

Prices and included usage

Pricing checked

Dify Cloud is billed per workspace; included message credits and allowances vary by plan. The self-hosted Community Edition has its own license terms.

Sandbox
USD 0

One member, five apps, 200 message credits, 50 knowledge documents.

Price source
Professional
USD 590/workspace/year; applicable taxes extra

Three members, 50 apps, 5,000 message credits/month.

Price source
Team
USD 1,590/workspace/year; applicable taxes extra

50 members, 200 apps, 10,000 message credits/month.

Price source
Community Edition
USD 0 software fee

Self-hosting within the Dify Open Source License.

Price source
  • Self-hosting adds infrastructure and, where applicable, model costs. OpenAI's price list is one example of separately billed model access: https://developers.openai.com/api/docs/pricing
  • Multi-tenant operation requires commercial authorization under the published license; visible branding is subject to additional conditions.

Pricing details

05

Sources and evidence

Official Dify documentation and repository license reviewed on 9 October 2026. Classic and new Agent nodes have different capabilities. Examples are editorial; no installation was tested here.

Editorial team: Michael Meierhoff · sources reviewed on . Provider information is not independent evidence of performance.

Read profile as Markdown

How we assess the sources