Releases: chuanqq/user-prompt-optimizer
Releases · chuanqq/user-prompt-optimizer
Release list
Release v0.6.0
What's new
- Role injection (optional): when enabled in settings, optimized prompts now begin with a domain role line (e.g. "You are a senior backend specialist, facing the following task/requirement:") so the optimized prompt is ready to hand to another model. The domain is auto-inferred from the prompt's subject matter, or a fixed domain hint can be pinned. Works in both template and lite modes.
Off by default — existing behavior is unchanged.
0.5.0
0.4.0
What's Changed
- Idle-timeout semantics: the request timeout clock now resets on every streamed chunk, so long reasoning generations are no longer cut off mid-stream (previously a total-duration timeout from request start)
- Settings UI: new "Request timeout" field under a "Request" section (seconds; 0 = no limit; default 60)
- Blank timeout input now keeps the current value instead of silently disabling the timeout
0.3.1
- Address Obsidian reviewer source-code warnings (W2-W11)
- Keep streaming-required fetch (with explanatory comments)
- Improve type safety in SSE JSON parsing
- Use Obsidian DOM helpers (createEl / setCssStyles / setCssProps / Setting.setHeading)
- Wrap async event callbacks for void-returning signatures