Versions and Upgrading

0.2.0 (2026-10-16)

Tag v0.2.0, release notes.

• A 429 that names one model's own quota (the free tier gives gemini-3.8-flash 20 requests a day) moves to the next model, which is then used for the rest of the run. The run stops with exit 3 only when every model's quota is spent, or at any other 429.
• Nothing else changed: same commands, same JSON record, same exit codes. Example outputs in this wiki were made with 0.1.0; apart from the version line they are the same in 0.2.0.

Upgrading: update the plugin (claude plugin update fetchwright@<marketplace>). If you put the slower model first to save the other's quota, keep doing so with --models or FETCHWRIGHT_MODELS.

0.1.0 (2026-10-15)

The first release. Tag v0.1.0, release notes, CHANGELOG.

There is nothing to upgrade from. Later releases will list here what changed in the commands, the JSON record and the exit codes, with the steps to move across.

Things likely to change

These depend on Google, not on fetchwright, and were true on 2026-10-15:

• The default models. gemini-3.8-flash and gemini-3.5-flash-lite. Set your own with --models or FETCHWRIGHT_MODELS without waiting for a release. The paid price of gemini-3.8-flash doubles on 2027-01-01, per Google's pricing page.
• The response shape. The Interactions API's steps (url_context_call, url_context_result, model_output) and the usage fields. A change there shows as pages "not read" with an unparsed answer, or as missing token counts.
• Retrieval statuses. Anything new is listed under new_statuses.