| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-08-14 | 1.6 kB | |
| v0.2.0 source code.tar.gz | 2026-08-14 | 22.4 MB | |
| v0.2.0 source code.zip | 2026-08-14 | 22.4 MB | |
| Totals: 3 Items | 44.8 MB | 0 | |
Added
- Let the shared Python vision client call either OpenAI Chat Completions or Responses APIs, including optional reasoning effort and explicit
store: falsedata handling. - Add native Anthropic Messages requests with protocol-specific authentication, image sources, optional thinking control, and text-block response extraction.
- Rewrite OpenAI Chat Completions
image_urlblocks through the existing vision-description pipeline with a host-neutral channel note. - Add a fast UI restoration workflow for quick first drafts while preserving page hierarchy, visible text, and native controls.
- Add the project landing-page source and link the DeepSeek Harness vision bundle for users who want that integration.
Changed
- Make upstream egress explicit and resilient: connect directly by default, optionally use a configured HTTP CONNECT proxy, and fail over only when connection establishment fails instead of following ambient system proxy settings.
- Make protocol-specific environment configuration authoritative while keeping existing host authentication and model compatibility behavior intact.
Fixed
- Send a browser-compatible, configurable User-Agent so Cloudflare-backed OpenAI-compatible endpoints do not reject Python's default signature.
- Honor
Retry-Afterand retry Anthropic 529 overload responses. - Run the
glancelauncher through its interpreter on Windows and preservetraceSVG byte output across Windows line-ending behavior.
Full Changelog: https://github.com/Anionex/agent-vision-toolkit/compare/v0.1.0...v0.2.0