Headless Session Runner FAQs

Question 1

Does it work within an existing Claude Code session?

Accepted Answer

Yes, the script automatically manages environment variables like CLAUDECODE to ensure that a background session can be successfully spawned from within an active Claude Code environment.

Question 2

When should I use Headless Runner instead of a standard subagent?

Accepted Answer

Use Headless Runner for larger implementation tasks where you want to dispatch work to the background with minimal context cost and receive a structured summary rather than manual turn-by-turn interaction.

Question 3

Which Claude models are supported?

Accepted Answer

It supports Haiku for simple mechanical tasks, Sonnet for standard feature implementation, and Opus for complex architectural changes or deep debugging.

Question 4

How does this skill handle Claude context limits?

Accepted Answer

The skill tracks per-turn token usage and provides a 'context_warning' flag when usage exceeds 60%, signaling that you should start a new session instead of resuming the current one.

Question 5

Can I see what happened during the background session?

Accepted Answer

Yes, the runner filters the JSON stream into a compact progress view and returns a comprehensive JSON object including exit codes, cost, duration, and a list of files changed via Git.

Headless Session Runner

주요 기능

사용 사례

Headless Session Runner

주요 기능

사용 사례