Forked from ceveyne/user-docs
Forked from ceveyne/user-docs
docs / CHANGELOG.md
docs / CHANGELOG.md
Notable changes to this project will be documented in this file.
annotate_image running a fresh detection on an annotated copy instead of the original. A target is now always looked up in the chat's media state first, however it is given (aN/iN/pN/vN, the original filename or the preview filename), and a result of detect_object, annotate_image or mask is never used as the source while its original is known.memorize_doc and rewrite_doc can now write and update Agent Skills–compatible SKILL.md skill bundles (with companion files) in the Notes Directory, in addition to plain notes.analyse_image can now read full generation metadata (prompt, model, sampler, steps, seed, LoRAs) from pictures sourced from a Draw Things project file, not just from saved images.annotate_image for photos with EXIF orientation metadata (e.g. phone camera photos).detectLabel) in annotate_image now works reliably when passing a list of labels.pN indexes.read_doc now returns readable text for PDF files instead of binary gibberish..rag-data) and the managed RAG Python environment (.rag-venv) to ~/.user-docs, so plugin updates no longer remove the local index or virtual environment.export_doc now copies LM Studio attachment previews from the chat working directory into images/ and rewrites the exported Markdown links to those preview files, so exported attachment images render reliably.annotate_image tool-result JSON no longer echoes the caller-supplied detection task; detection state and audit metadata still keep the internal task where needed.analyse_image and annotate_image now uses only the configured embeddingBaseUrl / LM-Studio-compatible Vision API path; the legacy local Python/FastVLM backend was removed.Notable changes to this project will be documented in this file.
annotate_image running a fresh detection on an annotated copy instead of the original. A target is now always looked up in the chat's media state first, however it is given (aN/iN/pN/vN, the original filename or the preview filename), and a result of detect_object, annotate_image or mask is never used as the source while its original is known.memorize_doc and rewrite_doc can now write and update Agent Skills–compatible SKILL.md skill bundles (with companion files) in the Notes Directory, in addition to plain notes.analyse_image can now read full generation metadata (prompt, model, sampler, steps, seed, LoRAs) from pictures sourced from a Draw Things project file, not just from saved images.annotate_image for photos with EXIF orientation metadata (e.g. phone camera photos).detectLabel) in annotate_image now works reliably when passing a list of labels.pN indexes.read_doc now returns readable text for PDF files instead of binary gibberish..rag-data) and the managed RAG Python environment (.rag-venv) to ~/.user-docs, so plugin updates no longer remove the local index or virtual environment.export_doc now copies LM Studio attachment previews from the chat working directory into images/ and rewrites the exported Markdown links to those preview files, so exported attachment images render reliably.annotate_image tool-result JSON no longer echoes the caller-supplied detection task; detection state and audit metadata still keep the internal task where needed.analyse_image and annotate_image now uses only the configured embeddingBaseUrl / LM-Studio-compatible Vision API path; the legacy local Python/FastVLM backend was removed.