IronClaw Setup and First Run

1. What installation actually gives you

The current upstream README highlights the Reborn runtime as the standalone path to focus on. The first meaningful commands are repo-local:

cargo run -q -p ironclaw_reborn_cli --bin ironclaw-reborn -- --help

or:

cargo build -p ironclaw_reborn_cli --bin ironclaw-reborn
./target/debug/ironclaw-reborn --help

2. The default home and isolated state

The default Reborn home is:

$HOME/.ironclaw/reborn

The README also shows how to override it for isolated testing:

export IRONCLAW_REBORN_HOME="$PWD/.reborn-home"

That is a good pattern for evaluation because it keeps trial state separate from anything more permanent.

3. Configure the model route

The upstream docs show a CLI-native setup flow:

ironclaw-reborn models set-provider openai --model gpt-5-mini

Then verify with:

ironclaw-reborn models status
ironclaw-reborn models list openai

This is a good example of IronClaw's philosophy: configuration is explicit, inspectable, and commandable.

4. What a successful first run looks like

A good first run should prove:

1. the home path is correct,
2. one model route resolves,
3. one secret is provided through the environment,
4. run or repl works without unclear side effects.

5. Good evaluation flow

Use this order:

1. config path and doctor,
2. set one provider,
3. export one API key environment variable,
4. test run --message,
5. try repl only after the basic route works.