v1.7.4

Mintplex-Labs/anything-llmv1.7.4Feb 18, 2025by timothycarambat

AI Summary

A significant feature release introducing the no-code agent skill builder and local OCR scanning for PDFs and images. The release also synchronizes Docker and Desktop versions.

Key Highlights

  • No-code agent flow builder is live
  • Local OCR scanning for PDFs and images (via Tesseract)
  • Synced Docker and Desktop release versions
  • Thinking/agent animation and UI improvements
  • Turkish language support

New Features

  • No-code agent flow builder
  • OCR support for images and non-digitized PDFs
  • Turkish translation
  • Confluence personal access token support
  • Agent flow builder move block fix

Full Release Notes

## The no-code agent skill builder is live ⚡ 
![AnythingLLM](https://github.com/user-attachments/assets/14b09cc2-4932-47c1-a81d-5b96ab64504a)

## Big changes

### 🏗️ The built-in no-code agent flow builder is alive!

Now, without a line of code you can seamlessly draft and run agentic skills that _just make sense_. No confusing setup - just AI skills on steroids, fully local and about 1000x easier to understand than MCP.

[Read the docs on making your first skill ->](https://docs.anythingllm.com/agent-flows/tutorial-hackernews)

### OCR scanning of PDF and images is live!

Now, when Anythingllm is given an image or _nondigitized_ PDF we will automatically OCR the text content into a document. This is done via a local [Tesseract](https://github.com/tesseract-ocr/tesseract?tab=readme-ov-file#installing-tesseract) model running in process!

_this is OCR, not "vision" so if the content has no text - nothing will be found!_


Lastly, we skipped a few Docker versions to make Docker <=> Desktop the same release version. It was confusing have two different semantic versions. They will now be lock-step going forward.

## All the other fixes 🔧 
* Enable `include_reasoning` in request by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3067
* parse streams from `reasoning_content` from DeepSeek API by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3068
* Add tokenizer improvements via Singleton class and estimation by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3072
* autodetect parseable text file contents by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3079
* inherit font size from UI for thoughts by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3099
* Enable consistent styling on chart items by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3126
* Add reasoning flag for azure models with "default" fallback by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3128
* Enable `num_ctx` to match defined chunk length in ollama embedder by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3129
* Patch PPLX streaming for timeouts by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3130
* Add improved logging and introspection to web-browsing by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3140
* [FEAT] Add Turkish Language Support (tr.js) #3108 by @TunahanDilercan in https://github.com/Mintplex-Labs/anything-llm/pull/3141
* [CHORE] Onboarding language translations by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3176
* fix UserMenu rendered twice on Main page by @sdeleeuw in https://github.com/Mintplex-Labs/anything-llm/pull/3110
* Add handling to reasoning models for Generic OpenAI connector by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3183
* Agent flow builder by @shatfield4 in https://github.com/Mintplex-Labs/anything-llm/pull/3077
* chore: rename Github to GitHub by @setchy in https://github.com/Mintplex-Labs/anything-llm/pull/3199
* chore: rename `Gitlab` to `GitLab` by @setchy in https://github.com/Mintplex-Labs/anything-llm/pull/3200
* Add `querySelectorAll` capability to web-scraping block by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3186
* Turkish translations by @cobanov in https://github.com/Mintplex-Labs/anything-llm/pull/3207
* OCR PDFs as fallback during upload by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3204
* Add OCR of image support by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3219
* feat(dataconnectors): support confluence personal access token by @lcgash in https://github.com/Mintplex-Labs/anything-llm/pull/3206
* fix agent builder move block bug by @shatfield4 in https://github.com/Mintplex-Labs/anything-llm/pull/3220
* Drop Node `canvas` for manual `sharp` conversion by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3221
* for projects in gitlab subgroup (#3075) by @t2tx in https://github.com/Mintplex-Labs/anything-llm/pull/3247
* Fix: fix text and format issues in Chinese documentation. by @alongLFB in https://github.com/Mintplex-Labs/anything-llm/pull/3249
* fix: sanitizeNamespace by @hakeemsyd in https://github.com/Mintplex-Labs/anything-llm/pull/3246
* fix: Patch agent flow to honor agent handler established provider by @timothycarambat in https://github.com/Mintplex-Labs/anything-llm/pull/3251

## New Contributors
* @TunahanDilercan made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3141
* @sdeleeuw made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3110
* @setchy made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3199
* @cobanov made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3207
* @lcgash made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3206
* @t2tx made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3247
* @alongLFB made their first contribution in https://github.com/Mintplex-Labs/anything-llm/pull/3249

**Full Changelog**: https://github.com/Mintplex-Labs/anything-llm/compare/v1.4.0...v1.7.4