AI LLM Search 1.1.5

What changed

• Search selected topics instead of generic pronouns such as “这是什么”.
• Filter unrelated web candidates, including exact acronym matching for cases such as NAIC versus NAICS.
• Group direct context, vault retrieval, and web candidates so users can see exactly what was sent to the model.
• Improve source titles, paths, hostnames, and narrow-sidebar readability.
• Add regression tests for context-query construction and source relevance.

AI LLM Search 1.1.4

Fixes selected-text context delivery for OpenAI-compatible models that only reliably honor the first system message. Web Viewer selections, current-note context, vault results, and web sources are now attached directly to the current user turn while the visible chat remains concise.

AI LLM Search 1.1.3

Fixes DeepSeek API compatibility after the OCR rollback. Bare https://api.deepseek.com addresses are now normalized to /v1/chat/completions, and the connection test again uses the original working request parameters. OCR remains removed; Web Viewer selected-text actions remain available.

AI LLM Search 1.1.2

Rollback

• Remove PDF/image OCR capture and all vision API settings.
• Restore text-only model requests.
• Keep selected-text actions inside Obsidian’s built-in desktop Web Viewer.
• Add a bounded connection test so the settings button cannot wait indefinitely.

AI LLM Search 1.1.1

Compatibility update

• OCR can now use a separate OpenAI-compatible API address, model, and SecretStorage key.
• Regular note chat can keep using DeepSeek while PDF/image OCR uses a multimodal provider.
• Blank vision API fields continue to reuse the main provider.

AI LLM Search 1.1.0

New

• Select native PDF text and ask AI from the context menu.
• Drag-select regions in scanned PDFs and images for multimodal OCR questions.
• Ask about selected text inside Obsidian’s built-in desktop Web Viewer, with the page URL preserved as context.
• Configure an optional dedicated vision/OCR model.

Privacy

OCR sends only the selected crop, and only after you press Send. API keys remain in Obsidian SecretStorage.

AI LLM Search 1.0.3

Community review fixes:

• Adopt Obsidian 1.13 searchable declarative settings.
• Preserve user-positioned sidebar leaves across reloads.
• Exclude the configured vault settings directory from retrieval.
• Handle asynchronous UI actions explicitly and remove deprecated APIs.
• Build and attest release assets with GitHub Actions.

AI LLM Search 1.0.2

Community-submission build under the AI LLM Search name and ai-llm-search plugin ID. Includes selected-text, current-note, whole-knowledge-base BM25 retrieval, optional web search, OpenAI-compatible model APIs, SecretStorage integration, and responsive desktop/mobile layouts.

Vault Context AI 1.0.1

Initial community-submission build under the unique Vault Context AI name and vault-context-ai plugin ID. Includes selected-text, current-note, whole-vault BM25 retrieval, optional web search, OpenAI-compatible model APIs, SecretStorage integration, and responsive desktop/mobile layouts.

Vault AI Assistant 1.0.0

Changelog

1.0.0

• Add current-note, selected-text, whole-vault, and web-assisted chat.
• Add local BM25 vault retrieval with source links.
• Add reading-view and editor context-menu actions for selected text.
• Add OpenAI-compatible API and SecretStorage configuration.
• Add responsive desktop and mobile sidebar layouts.