Installation

Prerequisites

Docker (for tama brew)
An API key for at least one supported LLM provider

Homebrew (macOS / Linux)

brew install mlnja/tap/tama

Build from source

Requires Rust 1.76+.

git clone https://github.com/mlnja/tama
cd tama
cargo build --release

This produces two binaries:

target/release/tama — developer tool
target/release/tamad — runtime

Add them to your PATH:

export PATH="$PATH:$(pwd)/target/release"

Verify

tama --help
tama run --help

Environment variables

tama uses environment variables to configure LLM providers and model roles.

API keys

Variable	Provider
`ANTHROPIC_API_KEY`	Anthropic (Claude)
`OPENAI_API_KEY`	OpenAI (GPT-4, etc.)
`GEMINI_API_KEY`	Google (Gemini)

Model roles

tama uses a role-based model system. Instead of hardcoding a model name in each agent, you assign roles (like thinker, writer, fast) and map them to models at runtime:

export TAMA_MODEL_THINKER="anthropic:claude-opus-4-6"
export TAMA_MODEL_WRITER="anthropic:claude-sonnet-4-6"
export TAMA_MODEL_FAST="anthropic:claude-haiku-4-5"

Agents reference roles:

call:
  model:
    role: thinker

This lets you swap models without editing any agent files.

Supported providers

Provider	Format	Example
Anthropic	`anthropic:model-id`	`anthropic:claude-sonnet-4-6`
OpenAI	`openai:model-id`	`openai:gpt-4o`
Google	`google:model-id`	`google:gemini-2.0-flash`