EN / ES
BYOK • Multi-Model • Local AI Ready

One IDE. Every Model.

Connect OpenAI, Anthropic, Google Gemini, OpenRouter, and Ollama in a single agentic development environment. Bring your own API keys or run models locally.

Agent14 IDE Workspace

Integration

Use the Models You Already Pay For

OpenAI Anthropic Claude Google Gemini OpenRouter Ollama

Agent14 connects directly to your existing providers.

Bring Your Own Key

Pay only for the tokens you use. Plug in your API keys for OpenAI, Anthropic, Gemini, or OpenRouter.

Multi-Provider Workflows

Swap models on the fly. Use Claude 3.5 Sonnet for planning and local models for fast compiling.

Local AI Support

Keep your code 100% private. Integrate Ollama to run advanced developer agents locally.

No Vendor Lock-In

You control the intelligence. Upgrade models or swap providers as soon as new LLMs are released.

Guidelines

Project-Aware Agent Memory

Agent14 automatically detects and reads custom guidelines to retain domain-specific knowledge across runs.

agents.md: Defines custom profiles, behaviors, and instructions for how the agent should write code.

skills.md: Logs reusable technical steps, compiler rules, and workspace boundaries to reduce context load.

Agent Guidelines and Skills Memory

Workflows

Built-In Planning and Task Tracking

Keep long-running tasks aligned. Agent14 generates, updates, and tracks files to document the engineering state.

plan.md: Dynamic implementation plan detailing files to be written, modified, or verified.

task.md: Live checklist of task progress, allowing developers to verify which sub-tasks have compiled successfully.

Task Management Checklist Workflow

Agent Customization

Control How Your Agent Works

Take command of agent behaviors via the built-in slash menu. Control reasoning levels, explanation styles, and execution rules directly.

Reasoning Visibility: See exactly what the agent is thinking. View raw execution steps, code search queries, and files mapped before writing.

Technical & Explanatory Modes: Toggle chat density. Technical mode spits pure diffs; Explanatory mode guides you through the architecture.

Silent Execution Mode: Let the agent work in the background. Disable chat narration to execute batch refactors without prompt interruptions.

Agent Customization Settings Interface

Code Integrity

Review Every Change

Every single code modification generates an interactive diff. You maintain absolute control over the codebase.

Granular Review

Accept or reject individual code hunks inside files. Approve complete batches after compiler passes.

Batch Commits

Commit changes directly when compilations are verified. Discard execution batches cleanly with a single click.

Git Diff Review Panel

UI Development

Work with Visual Context

Annotate your UI mockups directly in the IDE. Drag and drop screenshots, highlight specific layout areas, draw annotations, and add visual instructions that the agent parses to generate frontend components.

Visual Context Annotation Panel

Execution Engine

Autonomous Software Engineering Engine

Underneath the hood, Agent14 compiles, runs tests, and validates code integrity continuously.

AST Analysis & dependency Graphs

Builds abstract syntax trees to trace class structures, variables, and imports across all code paths.

Closed-Loop Test Runs

Launches test suites automatically. Diagnoses test failures via console output and patches code autonomously.

Workspace Safety Boundaries

Strict path verification protects the host operating system, securing files outside the project root.

Decision Engine

Choose Your Level of Autonomy

Switch between controlled intervention and fully autonomous code execution based on the task complexity.

Plan Mode

The agent proposes an implementation strategy before writing any files. Best for review loops.

Act Mode

The agent autonomously edits code, executes terminal tasks, and self-corrects until completion.

Built for Modern Software Stacks

Agent14 natively understands idiomatic conventions, build tools, and testing frameworks across systems languages.

BYOK Providers Local Models Terminal Execution Agent Workflows