v1.0.0

Episteme v1.0.0 — Release Notes 🔬

The Universal Research Intelligence & Truth Verification Layer

Episteme has been updated with critical enhancements across the FastAPI backend, the React sidebar browser extension, and the main landing page. These changes improve overall pipeline robustness, solve cross-origin security restrictions, and expand the analytical dashboard capabilities.

---

🚀 Key User-Facing Updates

1. Enhanced Extension Header Actions

• GitHub Integration: Added a new GitHub Codebase shortcut button directly to the sidebar extension header actions, allowing researchers and developers to view the project's codebase or download the packaged extension zip in one click.
• Theme Stability & Transitions: Resolved layout flashes during dark/light mode toggles on the landing page by syncing page header actions, HTML/body CSS classes, and establishing smooth transitions for color variables.

2. High-Contrast Terminal Text

• Visibility Fix: Staged a visual fix for the interactive console logs. Hardcoded high-contrast light text colors for terminal logs when operating in the light theme, ensuring complete visibility across all display settings.

3. Secure Jargon Explainer API Routing

• CSP & CORS Bypass: To bypass restrictive Content Security Policies (CSP) and cross-origin iframe rules enforced on secure journals (like Nature, IEEE, and local PDF viewers), the jargon highlighting explainer (/api/explain) now dynamically delegates network requests through the extension's background service worker (background.js).

---

⚙️ Developer & Backend Infrastructure Updates

4. Dynamic Robust Mock Mode (NVIDIA NIM Fallback)

• Failsafe Offline Workflows: When the backend detects that NVIDIA_API_KEY is not present in the environment variables, it no longer breaks or falls back to static hardcoded text. Instead, it fires a custom rule-based dynamic claim extractor (extract_dynamic_claims_from_text) and title-aware parser (get_dynamic_llama_mock) to feed authentic-looking claims and data structures to the LangGraph pipeline.
• Domain-Aware Video Feeds: Integrated automatic domain categorizations for related educational content. If a paper is GNN/graph-focused, it fetches Stanford CS224W tutorials; if it is transformer/LLM-focused, it recommends Karpathy and 3Blue1Brown walkthroughs; for diffusion/vision papers, it pulls Computerphile and Stanford vision modules.

5. Multi-Agent Data Schema Expansion

• Advanced Metrics Dashboard: Upgraded the FastAPI response models and the LangGraph graph output structure to fully compute and return:
  • Peer Review Report: Auto-generated structured evaluations highlighting strengths, weaknesses, recommendation, and questions for authors.
  • Evolution Timeline: Interactive mutations mapping the paper's theoretical ancestors and descendants.
  • Complexity & Readability Gauge: Custom scores estimating difficulty, reading time, and mathematical density.
  • Replication Repositories: Code registries containing links, Docker verification parameters, stars, and languages.
  • Author Network: Author affiliation indices and historical impact networks (h-index).

6. Caching Verification System

• Cache Completeness Enforcement: The API router (backend/app/main.py) now validates the integrity of cached data. If a cached analysis exists but lacks the newly implemented metrics (such as the timeline or author networks), the server automatically treats it as a partial cache miss and re-triggers the LangGraph pipeline to update the paper memory profile.

---

🛠️ Codebase Fixes

• Missing Import Resolution: Corrected a NameError: name 're' is not defined crash inside backend/app/pipeline/nodes.py by importing Python's regex module (re) at the top of the file, restoring functionality to the Conflict of Interest (COI) and reproducibility scanner nodes.