chore(release): prepare 0.6.0

This commit is contained in:
Patrick Wozniak
2026-08-25 18:06:46 +02:00
parent b6f84b0eda
commit f2bfb38996
3 changed files with 12 additions and 3 deletions
+9
View File
@@ -2,6 +2,8 @@
## Unreleased
## 0.6.0 - 2026-08-25
- Allow switching from a vision-capable model to a text-only model by omitting historical image tool results while preserving their text output; direct image prompts still fail clearly.
- Stream incremental tool-call arguments from the `/alpha/generate` transport instead of waiting for the final complete tool-call event.
- Add a daily GitHub Actions synchronization job that opens or updates a pull request for CLI version, image capability, reasoning, effort, and output-limit changes in the latest published Command Code catalog.
@@ -23,6 +25,13 @@
- Add isolated live E2E profiles for separate Go-, GOAT-, and Provider-plan credentials, covering transport selection, reasoning across turns, quota identity, aborts, tools, GOAT vision, Go image rejection, and packed-package validation.
- Fix extension load failure on newer pi hosts that reject registering a custom API under a built-in name (`openai-completions`); register under `commandcode-custom` instead and restore the real wire API before native compat dispatch.
### Contributors
- @jagaliano — added the live quota dashboard and hardened its integration.
- @omariqbalnaru — fixed custom API registration for Oh My Pi 17.4.0.
- @ThomasByr — added GLM-5.3 pricing and reasoning levels.
- @newCman1 — added DeepSeek V4 vision model support.
## 0.5.1 - 2026-08-11
- Add model-specific image input capabilities from the `command-code@1.15.1` catalog and forward user and tool-result images using the current Command Code wire format.