New in ABP Studio · AI Coding Agent

Ship ABP applications with an AI agent that understands ABP.

The ABP AI Coding Agent lives inside ABP Studio and works across your solution: planning, editing, building, running migrations, diagnosing runtime errors, and keeping every change Git-safe.

Try it in ABP Studio
Agent / Plan / Ask modes ABP docs subagent Semantic code search Git snapshot revert
ABP AI Coding Agent Overview

More than a generic coding assistant.

Cursor, Windsurf, Claude Code and Antigravity are strong general-purpose AI coding tools. ABP AI Coding Agent is positioned differently: it is framework-native for ABP teams.

ABP AI Coding Agent Positioning
Understands ABP solution structure
Modules, packages, entities, services, DTOs, DbContexts, permissions, localization resources and run profiles.
Acts across the delivery lifecycle
Code changes, builds, proxy generation, EF Core migrations, app startup, Docker containers and runtime diagnostics.
Designed for controlled autonomy
AI scopes, workflows, project rules, permission prompts, .abpignore and Git snapshot rollback.

Everything an ABP team needs in one agent workflow.

From first question to safe implementation, the agent works with the same ABP Studio context your team already uses.

01

Multi-session chat

Run multiple agent sessions as tabs, keep conversations persistent, restore after restart, and receive notifications when background work completes.

02

Three work modes

Use Agent for autonomous execution, Plan for approval-first implementation plans, and Ask for read-only guidance.

03

ABP-native context

The agent sees your ABP modules, services, DTOs, permissions, localization and active run profile before it starts working.

04

Local code intelligence

Indexing and semantic code search help the agent find relevant code across large solutions and external folders.

05

Workflows and rules

Inject project conventions, shared team workflows, reusable skills, and learned lessons into future sessions automatically.

06

Runtime diagnosis

Read exceptions, logs, requests and distributed events from ABP Studio's monitor to debug real running applications.

07

Application control

Start and stop apps, containers and run-profile tasks, then inspect logs or return failed output directly to the chat.

08

ABP CLI automation

Generate C# and Angular proxies, install client libraries, add EF Core migrations, and target builds by module or package.

09

Git intelligence

Review diffs, generate Conventional Commit messages, send GitHub issue or PR context to the assistant, and revert agent turns with snapshots.

Choose the right level of autonomy.

Use the same agent for Q&A, planning and implementation without switching tools.

Ask Mode Preview

Ask Mode

Read-only answers for architecture questions, ABP usage, code exploration and troubleshooting guidance.

Ultimate AI Agent Control.

Customize every aspect of your AI workflow. From selecting the most powerful LLMs to defining strict operational rules, you are in the driver's seat.

Access the World's Best AI Models

Connect your workspace to top-tier LLM providers instantly. Choose between Claude, GPT, Gemini, or DeepSeek and build your core infrastructure with elite models.

Model Settings Dashboard

Define Strict Rules

Inject custom guardrails and precise instructions. Create specialized behavioral prompt rules that shape exactly how your agent handles files, code style, or communication.

Create Agent Rules Panel

Smart Task Routing

Keep your performance high and costs low. Assign heavy models for complex reasoning tasks, while automatically routing simpler tasks to lightweight, rapid models.

Task Routing Configurations

Import Dynamic Skills

Seamlessly upload your markdown skill libraries. Your AI will dynamically discover and fetch these skills only when required, protecting the core context scope.

Import Skills Interface

How it works inside ABP Studio.

The agent combines natural-language requests with ABP-specific context, local RAG, tools, runtime signals and Git safety.

ABP AI Coding Agent How It Works
1

Compile context

The prompt includes solution structure, selected AI scope, workflow, AI rules and active plan.

2

Research and plan

It can use semantic code search, ABP docs, web-search and read-only research subagents.

3

Execute with tools

Builds, migrations, proxies, containers, apps and tasks are handled from the same chat flow.

4

Verify and recover

Runtime monitor data and Git snapshot rollback make debugging and safe iteration easier.

From GitHub issue to pull request.

See the full delivery loop inside ABP Studio — issue intake, AI review, feedback and PR creation without context switching.

GitHub Issue Integration

Seamlessly pull targeted issues from GitHub directly into the AI Agent to kickstart automated development inside ABP Studio with a single click.

AI Code Review

Trigger automated, in-depth AI code reviews directly within your local workspace to rapidly identify anomalies, bugs, and architecture alignment.

Review Feedback Loop

Instantly route the generated AI review insights back into the AI Agent for frictionless, automated code refinement and iterative fixes.

Commit and Pull Request

Generate smart, context-driven AI commit messages, push your changes to GitHub, and open Pull Requests directly without ever leaving ABP Studio.

Prompt examples for real ABP work.

Use the agent like a teammate who understands ABP conventions and your solution structure.

// Feature implementation
Create a plan to add subscription billing to the SaaS module.
Include entities, permissions, application services, UI changes,
migrations, proxy generation and tests. Do not edit files until I
approve.
// Runtime debugging
Start the AuthServer and Web app, reproduce the failing login request,
inspect exceptions and logs, then propose the smallest safe fix.
// ABP-specific maintenance
Add an EF Core migration for the new OrderStatus field.
Detect the correct startup project, run the migrator task,
generate C# and Angular proxies, then build affected packages.
// Controlled refactoring
Use the Payments AI Scope only.
Refactor duplicated validation logic into a domain service.
Follow our team AI rules and create a Git-safe snapshot first.

Safe autonomy for professional teams.

AI agents are powerful because they can act. ABP Studio adds practical controls so teams can move fast without handing over the whole workspace blindly.

ABP AI Coding Agent Security Controls
.abpignore Use gitignore-style rules to block secrets, certificates, environment files and private keys.
Permission prompts Shell commands, URL fetches and downloads require approval with a human-readable reason.
Scope isolation Restrict sessions to named AI Scopes such as a module, package or external folder.
Git snapshot revert Each agent turn is backed by a lightweight snapshot so changes can be restored quickly.

Enterprise AI Trust & Data Handling

Not yet, but it is coming soon for Enterprise plans. Enterprise customers will be able to use bring your own key (BYOK) and connect OpenAI-compatible custom LLM providers, including organization-approved providers or self-hosted OpenAI-compatible endpoints. Until then, all plans use the models provided through ABP AI credits. See our pricing page for license options.

No! The Agent runs locally in ABP Studio. Your prompt and the code context it carries go from Studio straight to the LLM provider that serves the model you selected so your requests and responses are not transmitting over ABP or Volosoft servers. And the chat history Studio keeps is stored locally on your machine, per solution.

No! Your code never comes to ABP/Volosoft servers therefore we don't use your code for anything. Also we don't permit LLM providers to train their models with the context you provide.

When an LLM generates code, this is under your responsibility to check it against different topics including licensing. We provide you popular LLM providers in the market and you agree their terms and conditions. We suggest you always review the generated code for security, licensing, scalability and maintainability.

Credits are purchased and tracked per organization. Organization owners can see the full AI credit balance, purchase history, and usage of all members on the Organization Management page. Developers can see the organization's remaining credit balance and their own usage on the My Organization page; they cannot see other members' usage. When AI credits are running low or completely exhausted, a warning email is sent to the organization owner and all developers in the organization. Organization owners can configure the warning email threshold on the Organization Management page.

How ABP AI Coding Agent is different.

The goal is not to replace every AI coding tool. The goal is to give ABP teams an agent that works natively with ABP Studio and ABP Framework conventions.

Capability ABP AI Coding Agent Generic AI coding tools
ABP solution awareness Built in — modules, DTOs, permissions, localization, DbContexts and run profiles. Usually inferred from files or docs supplied as context.
ABP docs-first answers Built in — dedicated ABP documentation subagent. Often relies on general web search, project files or model memory.
ABP CLI actions Built in — proxies, install-libs, EF migrations and targeted builds. Possible through terminal commands, but not ABP-native by default.
Runtime monitor access Built in — exceptions, logs, HTTP requests and distributed events. Typically requires custom setup, terminal logs or external integrations.
Plan-driven ABP development Built in — persisted plan documents and live step states. Available in some tools, but not tied to ABP Studio plan storage.
Git-safe rollback per turn Built in — Git snapshot revert retained for short-term recovery. Varies by product; many rely on normal Git history/checkpoints.

“For developers, it reduces context switching. For product managers, it shortens the path from feature request to tested implementation plan.”

Why product teams should care.

ABP AI Coding Agent is not just a productivity layer for developers. It helps teams standardize implementation plans, preserve project knowledge, reduce repeated mistakes through learned lessons, and make delivery status easier to inspect.

Faster feature scoping Ask for affected files, technical risks and implementation steps before work starts.
Better consistency Shared workflows and AI rules encode team conventions directly into the agent workflow.
Lower delivery risk Builds, runtime checks, plan progress and Git snapshots create a safer feedback loop.

FAQ

Clear answers for developers, team leads and product managers.

No. ABP AI Coding Agent is positioned as an integrated agent experience inside ABP Studio, not as a separate IDE.

Yes, in Agent mode. Use Plan mode when you want an approval-first workflow, or Ask mode when you only need read-only answers and guidance.

It receives ABP-aware context before execution. The agent can work with ABP solution structure, search official ABP docs through a dedicated subagent, and use tools tailored to common ABP development tasks.

Yes. AI Scopes, .abpignore, project rules, workflows, permission prompts and Git snapshots help teams control context and risky operations.

Not necessarily. Those are broad AI coding tools. ABP AI Coding Agent is designed for ABP Studio users who want ABP-native context, workflow control and lifecycle automation.

Yes. You can use ABP Studio with any ABP version.

AI credits are purchased per organization. To buy AI credits, go to the organization page, select the organization you want to purchase credits for, and complete the credit purchase there.

AI credits purchase

Build ABP apps with an agent that knows the framework.

Start with a plan, approve the work, let the agent implement, build, diagnose and recover safely inside ABP Studio.

Download ABP Studio
1
ABP Assistant
🔐 You need to be logged in to use the chatbot. Please log in first.