Spaces:
Sleeping
Sleeping
Commit History
-
fix: proper rewards for all steps and clean CSV output
- Added rewards to plugin, planner, planner_python, navigator, navigator_python steps - Navigation step now gets 0.5 reward on success - Extraction step calculates reward based on repo count (0.5 per repo + 1.0 bonus) - CSV output now returns clean data without nested structure - format_output() checks for csv_output key and returns it directly - Total rewards now accumulate correctly across all steps13402be -
feat: enhanced step accordion UI with Lucide icons and test report
- Added StepAccordionItem component with action-specific icons - Icon mapping for navigate, extract, plugins, planner, verify, etc. - Color-coded step statuses (running, completed, failed) - Expandable step details showing extracted data - Steps summary in completed view - Created comprehensive test report (10/10 tests passed) - GitHub trending scraper working with CSV outputdc6c23a -
fix: intelligent navigation scraper with GitHub trending support
a04acb3 -
feat: implement intelligent agentic web scraper
- Added intelligent navigation planning based on instructions - Created GitHub trending repository scraper strategy - Enhanced planner to understand goals like 'trending repos' - Added CSV output for repository data (username, repo_name, stars, forks) - Improved agent orchestration for goal-oriented scraping - Added intelligent exploration vs single-page strategies The scraper now understands instructions like 'Get me all trending repo' and navigates intelligently to GitHub trending pages instead of just doing basic field extraction from the homepage.82fb385 -
fix: resolve scraper functionality and plugin issues
- Fixed plugin registry missing web_scraper and python_sandbox - Removed locals() from sandbox BLOCKED_CALLS for analysis - Fixed frontend health check API response parsing - Added comprehensive test validation framework - Verified all agents (planner, navigator, extractor, verifier) working - Confirmed Python sandbox execution with numpy/pandas/bs4 - Validated real-world URL scraping capabilities - Added session artifact management and memory integration54ec9cb
-
test: comprehensive ScrapeRL system tests - 100% pass rate
- Test scraper environment at LOW/MID/HIGH complexity - Test reward function with ground truth accuracy - Test plugin system installation/uninstallation - Test Gemini embeddings with similarity search - Test vector search/memory manager - Test NVIDIA and Groq AI providers - Test API endpoints (tasks, plugins, episode lifecycle) Components tested: Scraper, Reward, Plugins, Embeddings, Memory, AI Providers, API Total: 21 tests, 21 passed (100%)e123ba8 -
chore: move AI test report to docs/test
852ba36 -
test: comprehensive AI provider testing with 90% success rate
da142bc -
docs: comprehensive README update with all new features and examples
9160ee4 -
test: add end-to-end scraper test script with successful validation
8512126 -
feat: add frontend WebSocket integration with episode progress tracking
fa81e4d -
feat: add WebSocket support for real-time scraper progress updates
4afa792 -
feat: redesign UI with navy blue and cyan theme colors
05f6bf1 -
feat: add providers API endpoints with model and cost tracking
1226a86 -
feat: add embeddings service and update plugins with NVIDIA support
e5b4f8d -
feat: update model router with NVIDIA support and latest model scores
101ad87 -
feat: add NVIDIA provider and update Gemini/Groq with latest models
f5ba363 -
feat: add NVIDIA API key support and additional env vars
f080be2
-
docs: add manual test report with 89 tests
- Create MANUAL_TEST_REPORT.md with comprehensive UI tests - Test all Dashboard features (input view, dashboard view) - Test all popups (model, vision, agents, plugins, task type) - Test Settings, Plugins, Docs pages - Test API integrations and Docker - All 89 manual tests passed - Update TEST_REPORT.md to reference manual tests5608193 -
feat: redesign Dashboard with 2-view split and improved UX
Dashboard Input View (ChatGPT-like): - Multi-URL input with add/remove functionality - Structured instruction and output format fields - Model/Vision/Agents/Plugins/Type selection buttons - Clean centered layout with visual hierarchy Dashboard View: - Left sidebar shows only INSTALLED/ENABLED items - Info icons on each item with popup details - Session-based stats (start at 0, not fake data) - Assets section replaces Recent Actions - Memory management with add/view functionality - Input summary in right sidebar Additional changes: - Move Swagger docs from /docs to /swagger - Add vision model selection option - Add info popup component for item details - Fix system status displayde797a5 -
fix: support settings endpoint without trailing slash
f659d59 -
fix: support plugins endpoint without trailing slash
3232831 -
docs: update test report with build status
480ddb9 -
feat: redesign Dashboard and Settings with proper layout
Dashboard: - Add structured input bar with URL, instruction, task type - Add left sidebar with accordions (Agents, MCPs, Skills, APIs, Vision) - Add center area with stats header, main visualization, logs terminal - Add right sidebar with memory stats, extracted data, recent actions - Add popup modals for model/agent/plugin selection - Add task complexity selector (Low/Medium/High) Settings: - Add left sidebar navigation with sections - Add Budget & Limits section (disabled by default) - Add Appearance and Notifications settings - Add proper section-based layout App: - Add Docs tab in navigation - Remove footer, use full height layout - Compact navbar design Docs: - Add USER_GUIDE.md documentation - Add DocsPage component with react-markdown - Include agents, plugins, memory, API reference docs Types: - Add autoSave and debugMode to SystemSettings - Add AgentOption and ModelOption interfacesf881423 -
feat: add static file serving and improve UI/UX design
- Add static file serving in FastAPI for frontend assets - Add SPA routing support for React Router - Redesign App.tsx with modern gradient navbar and footer - Improve Dashboard with stat cards and cleaner layout - Enhance Settings page with 2-column standalone design - Redesign PluginsPage with category colors and better cards - Fix TypeScript lint errors (unused imports)039e1ea -
docs: update test report with 116 tests passing
5e3d245 -
test: add comprehensive API and core module tests
- Add 30+ new tests for memory, tasks, episode modules - Episode tests cover lifecycle, steps, manager operations - Memory API tests cover store, query, delete operations - Tasks API tests cover list, filter, create operations - All 101 tests passing with 44% coveragee8d7c11 -
fix: correct Docker healthcheck endpoint path
864b733 -
docs: update test report with latest results
fbb1d81 -
fix: add list_providers method and agents list endpoint
e497b8a
-
chore: add test_models package init
bf7d25d -
docs: add test report with coverage analysis
ccaf48d -
fix: replace deprecated datetime.utcnow with timezone-aware
bfe0e24 -
feat: register settings and plugins API routes
9852326 -
feat: add modular plugin system for extensions
942fc42 -
feat: add client API key configuration and model selection
5c03d80 -
fix: update docker-compose and package-lock for npm ci
0a8f919 -
test: add frontend test suite with vitest
ebe1c3d -
chore: update gitignore to exclude cache files
a3fb547 -
docs: update README with HuggingFace Spaces deployment
ac3ec82 -
chore: add Docker configuration for HuggingFace Spaces
df861a1 -
feat: implement React dashboard with components and hooks
0cfd364 -
chore: initialize frontend with React, Vite, and TailwindCSS
f4a05c4 -
test: add comprehensive backend test suite
46eecf4 -
feat: add API routes and utility modules
27cde0c -
feat: add MCP tool registry and search engine integration
afefaea -
feat: implement hierarchical memory system
bb3ee41 -
feat: add multi-model LLM router with providers
ca1fd98 -
feat: implement multi-agent system with coordinator
3bfb250 -
feat: add core RL environment models (observation, action, reward, env)
ab65628