Skip to main content
Tembo supports coding agents to solve your software engineering tasks. Each coding agent has different capabilities, models, and configuration options.

​
Available Coding Agents

Coding AgentModel ProvidersDefault ModelRule Files
Claude CodeAnthropicclaude-4-5-sonnetclaude.md
CodexOpenAIgpt-5-mediumagents.md
OpencodeAnthropicclaude-4-5-sonnetagents.md
AmpAnthropic (auto-selected)claude-4-5-sonnetagents.md
CursorAnthropic/OpenAI/Google/Cursor/XAIclaude-4-5-sonnetagents.md

​
Claude Code (Default)

Claude Code is Tembo’s default coding agent, powered by Anthropic’s Claude models. Supported Models:
  • claude-4-5-sonnet (Default) - Balanced performance and cost
  • claude-4-5-haiku - Fastest, lightweight reasoning
  • claude-4.1-opus - Maximum capability for complex tasks
  • claude-4-sonnet - Fast and efficient

​
Codex

Codex is a coding agent powered by OpenAI’s latest GPT-5 models with varying reasoning levels, allowing you to choose the right balance of speed and thoroughness. Supported Models:
  • gpt-5-medium (Default) - Balanced reasoning
  • gpt-5-minimal - Fastest, lightweight reasoning
  • gpt-5-low - Quick reasoning
  • gpt-5-high - Deep reasoning for complex problems
  • gpt-5-codex - Specialized for code generation tasks

​
Opencode

Opencode is an alternative coding agent supporting both Anthropic Claude models and OpenAI GPT models, providing flexibility in model selection. Supported Models:
  • claude-4-5-sonnet (Default) - Balanced performance and cost
  • claude-4-5-haiku - Fastest, lightweight reasoning
  • claude-4.1-opus - Maximum capability for complex tasks
  • claude-4-sonnet - Fast and efficient

​
Amp

Amp is a powerful coding agent powered by Anthropic’s Claude models. Amp uses smart model detection and currently uses Sonnet 4.5 as its main model.

​
Cursor

Cursor is a coding agent supporting Anthropic Claude models, OpenAI GPT models, Google’s Gemini, Cursor’s Composer, and XAI’s Grok through the Cursor CLI. Supported Models:
  • claude-4-5-sonnet (Default) - Balanced performance and cost
  • claude-4.1-opus - Maximum capability for complex tasks
  • gpt-5.1 - Latest GPT-5.1 model
  • gpt-5.1-codex - GPT-5.1 specialized for code
  • gpt-5.1-codex-high - GPT-5.1 Codex with high reasoning
  • gemini-3-pro - Google’s Gemini 3 Pro model
  • composer-1 - Cursor’s Composer model
  • grok - XAI’s Grok model for advanced reasoning

​
Choosing the Right Coding Agent

Different coding agents excel at different types of tasks. Here’s guidance on when to use each: For most tasks: Start with Claude Code (default). It provides the best balance of code quality, reliability, and cost-effectiveness for general software engineering work. For complex refactoring or architectural changes: Use Claude Code with claude-4.1-opus or Codex with gpt-5-high for deeper reasoning capabilities. For quick fixes and simple updates: Use Claude Code with claude-4-5-haiku or Codex with gpt-5-minimal for faster execution on straightforward tasks. For team consistency: If your team already uses a specific IDE or coding environment (Cursor, Amp, etc.), select the matching coding agent for familiar code patterns and style.

​
Selecting a Coding Agent

Coding Agents are specified using a coding agent key format: agentType:model Examples:
  • claudeCode:claude-4-5-sonnet
  • codex:gpt-5-medium
  • opencode:claude-4-5-sonnet
  • amp:claude-4-5-sonnet
  • cursor:claude-4-5-sonnet
  • cursor:gpt-5.1-codex
  • cursor:grok
If no coding agent is specified, Tembo defaults to claudeCode:claude-4-5-sonnet.

​
Choosing the Right Model

Different models offer different tradeoffs between speed, cost, and capability: For most tasks: Use the default models (claude-4-5-sonnet or gpt-5-medium)
  • Balanced performance for typical engineering tasks
  • Good reasoning capability with reasonable speed
  • Cost-effective for production use
For simple, straightforward tasks: Use lightweight models (claude-4-5-haiku or gpt-5-minimal)
  • Faster execution for simple bug fixes
  • Lower cost for high-volume operations
  • Best for well-defined, smaller scope work
For complex architectural work: Use advanced models (claude-4.1-opus or gpt-5-high)
  • Deep reasoning for system design decisions
  • Better handling of large, multi-file refactors
  • More thorough analysis of edge cases

​
Using Rule Files

Rule files allow you to provide coding agent-specific instructions and coding standards. All coding agents support tembo.md and other standard rule files in your repository root. Tembo will look for and concatenate instructions from all of the following rule files if they exist in your repository:
  • CLAUDE.md
  • tembo.md
  • .rules
  • .cursorrules
  • .windsurfrules
  • .clinerules
  • .github/copilot-instructions.md
  • AGENT.md
  • AGENTS.md
These files can include:
  • Coding standards and style guidelines
  • Project-specific conventions
  • Commands to run before or after certain operations
  • Context about your architecture and patterns
Learn more about rule files.

​
Switching Between Coding Agents

You can configure which coding agent to use for your organization or specific tasks through:
  1. Dashboard Settings - Set a default coding agent for all tasks
  2. API Configuration - Specify a coding agent key when creating tasks via API
  3. Integration Settings - Configure per-integration coding agent preferences
Contact your Tembo administrator or check the API documentation for details on coding agent configuration.