chore: upgrade base image to python 3.11-slim, force HF rebuild

fix: update Space URL in README, fix HF template display

chore: trigger fresh HF build after org transfer

fix: switch sentiment to PyTorch, install CPU-only torch in Dockerfile

fix: remove tensorflow and tf-keras — not used, replaced with torch for transformers

feat: add source type guide expander with scoring rationale and 6-type table

fix: add _repair_ost auto-repair, fix fallback model, improve LLM error display

fix: dark text in OST editor inputs, results page top spacing, upload filename guide

fix: enlarge goal banner, switch upload page to wide layout to prevent sidebar clipping

feat: remove goal suggestions, auto-detect file source type, fix sidebar tagline

feat: use real PNG logo extracted from PPTX instead of SVG approximation

fix: improve logo SVG, add logo to sidebar on all pages, fix title layout

fix: force light mode with base=light in theme config

feat: apply PPT design system — logo, colors, theme, CSS across all pages

fix: block weak goals (score<2), remove proceed-anyway buttons

fix: remove demo OST, add page guards (goal→upload→results flow)

fix: show warning when GROQ_API_KEY missing on goal page

fix: revert to docker SDK — streamlit 1.32 incompatible with Python 3.13 on HF

fix: add streamlit config for HF proxy, goal guidelines, context always visible

fix: clean README frontmatter — streamlit SDK, sdk_version 1.32.0

fix: add app_port 8501 to HF Spaces README config

Merge branch 'main' into feature/ost-editor-cleanup

fix: resolve README merge conflict

Update README.md

Update fallback LLM version in README

fix: remove app_file from Docker SDK README

chore: add Dockerfile, fix SDK to docker in README

Add app metadata to README.md

chore: add HF Spaces config to README

Fix formatting in team section of README

docs: add README + remove misleading 500-chunk caption

Merge pull request #32 from Lookus22/feature/ost-editor-cleanup

Merge pull request #31 from Lookus22/chore/v-b-step3-source-diversity-validation

fix: complete session_state initialisation in app.py

@

Merge branch 'main' of github.com:Lookus22/discovery-lens

chore(notebooks): V-B step 3 source_type_diversity validation

chore(gitignore): broaden data/synthetic exclusion

fix(extractor): use 'object' dtype in CSV column selection

chore: gitignore JTBD export JSON files

Remove _legacy_pages gitignore

Add data/synthetic to gitignore

q

Delete legacy pages, remove synthetic data, add data/synthetic to gitignore

Merge pull request #30 from Lookus22/feature/cache-resource-warmup

Merge pull request #29 from Lookus22/validation/JTBD

fix: add CSV support and pandas str dtype fix to export_jtbds.py and extractor.py

fix: replace llama-3.1-8b-instant fallback with llama-4-scout (TPM limit workaround)

fix: move legacy pages out of pages/ to clear sidebar before demo

fix: remove debug expander from goal.py before demo