Configuration, output, and local files

资料正文

§1 Configuration precedence

The CLI resolves configuration in this order:

Explicit --config.
TTS_OPEN_TOOLKIT_CONFIG environment override.
Project configuration.
User configuration.
Built-in defaults.

§2 Default paths

Platform or scope	Path
macOS user config	`~/Library/Application Support/tts-open-toolkit/config.json`
Linux user config	`${XDG_CONFIG_HOME:-~/.config}/tts-open-toolkit/config.json`
Windows user config	`%APPDATA%\tts-open-toolkit\config.json`
Project override	`<project>/.tts_open_toolkit/config.json`
Project history	`<project>/.tts_open_toolkit/history/`
Project skills	`<project>/.codex/skills/`

Run tts_open_toolkit doctor to print the exact resolved paths on the current machine.

§3 OAuth storage

Profile metadata is stored in ~/.tts_open_toolkit/config.json.
OAuth access and refresh tokens are not stored in that file.
macOS uses Keychain; Windows uses CurrentUser DPAPI.
Status, list, and doctor output expose only non-secret metadata.

§4 Machine-readable output

tts_open_toolkit auth status --json
tts_open_toolkit devapi list --json
tts_open_toolkit app list --json
tts_open_toolkit skill state --scope project --json

Use JSON for agent workflows and automation. Do not scrape the human-readable format when a JSON option exists.

§5 Deterministic environments

Set OPEN_TOOLKIT_SKIP_UPDATE_CHECK=1 in offline or deterministic CI smoke tests. Keep package updates explicit.

§6 Was this page helpful?

Was this page helpful?

Tell us what worked, what was unclear, or what should be added. TTS Open Toolkit Feedback Form