diff --git a/01 Projects/barclays-rag-report/Barclays Rag Report.md b/01 Projects/barclays-rag-report/Barclays Rag Report.md index 7774346..0d8c70b 100644 --- a/01 Projects/barclays-rag-report/Barclays Rag Report.md +++ b/01 Projects/barclays-rag-report/Barclays Rag Report.md @@ -1,32 +1,63 @@ --- name: "Barclays Rag Report" -client: "TBD" +client: "Barclays" status: active -tech: [] +tech: + - Python + - OpenAI Assistants API + - GPT-4o local_path: /Users/ai_leed/Documents/Projects/Oliver/barclays-rag-report deploy: url: -server: +server: local tags: - project created: 2026-05-10 --- ## Overview -> New project — fill in during first session. +Standalone Python CLI tool that benchmarks OpenAI Assistants against RAG knowledge bases. Built for Barclays to evaluate banner-creation assistants (Internal Banners, Social Posts, Display Banners, PPC) — generates test prompts from brand/compliance docs, runs them against the assistant N times in parallel, scores responses on quality/consistency/accuracy/completeness (1–10 via GPT-4o), and produces an HTML report with charts. ## Tech Stack -- **Frontend:** -- **Backend:** -- **Infrastructure:** +- **Language:** Python 3.8+ +- **AI:** OpenAI Assistants API (beta.threads), GPT-4o for question generation + evaluation +- **Libraries:** openai, pandas, numpy, matplotlib, rich, langchain, scikit-learn, docx2txt +- **Infrastructure:** None — local script, no Docker + +## Structure + ## Deployment -- **Local path:** `/Users/ai_leed/Documents/Projects/Oliver/barclays-rag-report` +- **Local path:** +- **Run:** +- **Env:** +- **Install:** ## Sessions +### 2026-05-10 – Create a CLAUDE.md file documenting the +**Asked:** Create a CLAUDE.md file documenting the codebase architecture, build commands, and testing procedures. +**Done:** Generated CLAUDE.md with common development commands, high-level architecture overview, and testing instructions for the Obsidian CLI codebase. + ## Change Log | Date | Requested | Changed | Files | |------|-----------|---------|-------| +| 2026-05-10 | Documentation | Created CLAUDE.md with build/lint/test commands and architecture overview | CLAUDE.md | +| 2026-05-10 | Init session | Created CLAUDE.md, filled project note | CLAUDE.md | ## Related +- [[wiki/tech-patterns/]] — OpenAI API patterns +%% +Structure (filled programmatically): +rag_test_app/ - main app directory + main.py - RAGTester class + cli.py - CLI entry point + requirements.txt + tests/ - pytest suite +TEST_TO_RUN/ - Barclays configs JSON + ZIP documents + +Deployment: +Local only, no server. cd rag_test_app && python cli.py --config-dir ../TEST_TO_RUN/ +Install: pip install -r requirements.txt +Env: OPENAI_API_KEY +%% \ No newline at end of file diff --git a/99 Daily/2026-05-10.md b/99 Daily/2026-05-10.md new file mode 100644 index 0000000..9a104a6 --- /dev/null +++ b/99 Daily/2026-05-10.md @@ -0,0 +1,10 @@ +--- +date: 2026-05-10 +tags: [daily] +--- + +## Session Log + +- 13:01 (2min) | `barclays-rag-report` + - **Asked:** Create a CLAUDE.md file documenting the codebase architecture, build commands, and testing procedures. + - **Done:** Generated CLAUDE.md with common development commands, high-level architecture overview, and testing instructions for the Obsidian CLI codebase.