> ## Documentation Index
> Fetch the complete documentation index at: https://docs.infercom.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Agentic Coding

> Run AI coding agents on Infercom's EU sovereign infrastructure. MiniMax-M2.7 delivers top-tier coding performance at 400+ tokens/sec with 192k context.

Agentic coding tools are AI assistants that can autonomously write, edit, and refactor code in your codebase. Unlike simple code completion, these agents understand context, make multi-file changes, and iterate on their work.

## Why Infercom for Agentic Coding

<CardGroup cols={2}>
  <Card title="EU Sovereign" icon="shield-check">
    Full GDPR compliance. Data stays in EU. No US CLOUD Act exposure.
  </Card>

  <Card title="High Performance" icon="bolt">
    MiniMax-M2.7: 400+ tokens/sec - 6x faster than Claude Sonnet.
  </Card>

  <Card title="Quality" icon="star">
    MiniMax-M2.7: Top-tier coding performance with 192k context window.
  </Card>

  <Card title="No Rate Limits" icon="infinity">
    No quota drama. Scale your development without interruption.
  </Card>

  <Card title="No Quantization" icon="microchip">
    Always full precision. No quality loss from model compression.
  </Card>
</CardGroup>

## Supported Tools

We document the most popular tools below, but **any tool that supports OpenAI-compatible or Anthropic-compatible APIs will work with Infercom**. Just point it to `https://api.infercom.ai/v1` with your API key.

### Terminal Tools

| Tool                                    | Type         | Best For                                   |
| --------------------------------------- | ------------ | ------------------------------------------ |
| [Aider](/en/agentic-coding/aider)       | Terminal CLI | Simple, git-integrated workflows           |
| [Codex CLI](/en/agentic-coding/codex)   | Terminal CLI | OpenAI's agentic assistant (Responses API) |
| [OpenCode](/en/agentic-coding/opencode) | Terminal TUI | Interactive coding sessions                |
| [Goose](/en/agentic-coding/goose)       | Terminal CLI | Block's autonomous agent                   |
| [Warp](/en/agentic-coding/warp)         | AI Terminal  | Modern terminal with AI features           |

### IDE Extensions & Editors

| Tool                                       | Type              | Best For                   |
| ------------------------------------------ | ----------------- | -------------------------- |
| [Blackbox AI](/en/agentic-coding/blackbox) | VS Code Extension | AI coding assistant        |
| [Cursor](/en/agentic-coding/cursor)        | AI IDE            | Full AI-native code editor |
| [Cline](/en/agentic-coding/cline)          | VS Code Extension | Autonomous coding agent    |
| [Kilo Code](/en/agentic-coding/kilocode)   | VS Code Extension | Open source coding agent   |
| [Continue](/en/agentic-coding/continue)    | VS Code/JetBrains | Open-source, customizable  |
| [Windsurf](/en/agentic-coding/windsurf)    | AI IDE            | Cascade agentic AI         |

### Advanced

| Tool                                                    | Type  | Best For                    |
| ------------------------------------------------------- | ----- | --------------------------- |
| [Claude Code + ccproxy](/en/agentic-coding/claude-code) | Proxy | Staying in Claude ecosystem |

## Recommended Models

**MiniMax-M2.7** is the recommended model for agentic coding:

* **Context window:** 192K tokens (largest in EU infrastructure)
* **Speed:** 400+ tokens/sec
* **Capabilities:** Reasoning, function calling, tool use

<Note>
  MiniMax-M2.5 (160k context) remains available but will be deprecated in a future release. New projects should use MiniMax-M2.7.
</Note>

## Configuration Patterns

Most tools support two configuration approaches:

### Single Model (Simple)

Use MiniMax-M2.7 for all tasks. This is the simplest setup and works well for most workflows.

```
Model: MiniMax-M2.7
```

### Plan/Execute (Advanced)

For complex tasks, many tools support separate models for planning and execution:

| Phase       | Model               | Purpose                                             |
| ----------- | ------------------- | --------------------------------------------------- |
| **Plan**    | Claude, GPT, Gemini | Top-tier reasoning for architectural decisions      |
| **Execute** | `MiniMax-M2.7`      | Fast code generation on EU sovereign infrastructure |

This keeps planning quality high while routing the bulk of code generation to Infercom's fast, EU sovereign infrastructure.

Tools that support this pattern:

* **Codex CLI:** `[profiles.plan]` and `[profiles.execute]` in config.toml
* **Cline:** "Use different models for Plan and Act modes" toggle
* **OpenCode:** `agent.plan` and `agent.build` in opencode.json

<Tip>
  Start with the single-model setup. Switch to plan/execute when working on complex architectural tasks or multi-file refactoring.
</Tip>

## Getting Started

1. [Get your API key](https://cloud.infercom.ai/apis) (includes free credit)
2. [Choose a tool](/en/agentic-coding/choosing-a-tool) based on your workflow
3. Follow the setup guide for your chosen tool

<Tip>
  New to Infercom? Start with [OpenCode](/en/agentic-coding/opencode) - full TUI experience with reasoning visualization.
</Tip>

## API Details

```
Endpoint: https://api.infercom.ai/v1
Model: MiniMax-M2.7
```

See [OpenAI compatibility](/en/features/openai-compatibility) and [Anthropic compatibility](/en/features/anthropic-compatibility) for API details.