[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-81038":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":14,"stars7d":15,"stars30d":16,"stars90d":14,"forks30d":14,"starsTrendScore":14,"compositeScore":17,"rankGlobal":9,"rankLanguage":9,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":9,"pushedAt":9,"updatedAt":37,"readmeContent":38,"aiSummary":39,"trendingCount":14,"starSnapshotCount":14,"syncStatus":13,"lastSyncTime":40,"discoverSource":41},81038,"eve-agent-v2-unleashed","JeffGreen311\u002Feve-agent-v2-unleashed","JeffGreen311","Eve Agent V2 Unleashed — local-first autonomous AI coding agent powered by Ollama",null,"Python",35,8,2,0,5,7,2.86,"MIT License",false,"main",true,[23,24,25,26,27,28,29,30,31,32,33,34,35,36],"agentic-ai","ai-agent","autonomous-agent","chatbot","coding-assistant","fastapi","gpu","llm","local-ai","ollama","open-source","python","streaming","windows","2026-06-12 02:04:10","\u003Cdiv align=\"center\">\n\n\u003Cimg src=\"web\u002Fassets\u002Feve_v2u_logo.png\" width=\"220\" alt=\"Eve V2 Unleashed\" \u002F>\n\n# ◈ EVE AGENT V2 UNLEASHED ◈\n### Local soul. Cloud muscle. 40-round autonomous loop.\nYour GPU runs the personality. Qwen3 Coder 480B handles agentic heavy lifting via Ollama cloud.\nMid-loop complexity escalation — local model drives until the task needs frontier power.\nNo accounts. No telemetry. No per-token surprises.\n\n[![GitHub Stars](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FJeffGreen311\u002Feve-agent-v2-unleashed?style=for-the-badge&color=00ff41&labelColor=0c0c1a)](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed\u002Fstargazers)\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-00ffff?style=for-the-badge&labelColor=0c0c1a)](LICENSE)\n[![Python 3.11+](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPython-3.11%2B-ff00ff?style=for-the-badge&labelColor=0c0c1a)](https:\u002F\u002Fpython.org)\n[![Ollama](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPowered%20by-Ollama-ffe600?style=for-the-badge&labelColor=0c0c1a)](https:\u002F\u002Follama.com)\n[![Hugging Face](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F🤗%20Models-JeffGreen311-ff6600?style=for-the-badge&labelColor=0c0c1a)](https:\u002F\u002Fhuggingface.co\u002FJeffGreen311)\n\n**[🌐 My website](https:\u002F\u002Feve-cosmic-dreamscapes.com) · [🤗 Models](https:\u002F\u002Fhuggingface.co\u002FJeffGreen311) · [📦 Ollama Hub](https:\u002F\u002Follama.com\u002Fjeffgreen311) · [🐛 Report Bug](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed\u002Fissues)**\n\n\u003C\u002Fdiv>\n\n---\n\n## What Is Eve Agent V2 Unleashed?\n\nEve is an autonomous coding agent that **plans, executes, and verifies** multi-step programming tasks without hand-holding. She runs entirely on your local GPU through [Ollama](https:\u002F\u002Follama.com) — or optionally scales up to 480B cloud parameters when you need maximum firepower.\n\nThink **Claude Code**, but local-first, open-source, and built with a cyberpunk soul.\n\n```\nUser: \"Build me a FastAPI server with JWT auth and a PostgreSQL backend\"\n\nEve: [reads project] → [plans approach] → [writes 6 files] →\n     [runs tests] → [fixes 2 errors] → [verifies it works] → Done ✓\n```\n\n> **Original project** at [eve-cosmic-dreamscapes.com](https:\u002F\u002Feve-cosmic-dreamscapes.com) — Eve's full chat interface including Eve Coder and Eve Agent Portal.\n\n---\n\n## ✨ Key Features\n\n| | Feature | Details |\n|-|---------|---------|\n| 🔄 | **40-Round Agentic Loop** | Plans, executes, verifies, and self-corrects — up to 40 tool-call rounds per task |\n| ⚡ | **Real-Time Streaming** | Token-by-token SSE output — watch Eve think and build live |\n| 🛠️ | **Full Tool Suite** | bash, file I\u002FO, grep, glob, git, web search, URL fetch, multi-edit |\n| 🖥️ | **Local + Cloud Models** | Local GPU models AND Ollama cloud (480B) — switch mid-session |\n| 📁 | **Workspace Picker** | Change your working directory from the UI at any time |\n| 🤖 | **112 Sub-Agents** | Specialized agents for Python, FastAPI, Rust, ML, DevOps, security… |\n| 💬 | **111 Slash Commands** | `\u002Ffix`, `\u002Freview`, `\u002Frefactor`, `\u002Ftest`, `\u002Fdocs`, `\u002Fplan` and more |\n| 🧠 | **273 Skills** | Composable skill modules, progressively loaded |\n| 🔍 | **Live Web Search** | Tavily-powered — Eve researches the web mid-task |\n| 🗡️ | **Quest System** | Drop `.md` files into `workspace\u002Fquests\u002F` — Eve runs them automatically |\n| ⚡ | **RPG Progression** | Eve earns XP, levels up, and unlocks achievements as she works |\n| 📱 | **Telegram Bridge** | Push notifications and mobile chat with Eve |\n| 🪟 | **Windows Native** | PowerShell-aware bash tool, one-click `.bat` launcher |\n| 🎨 | **Cyberpunk UI** | Animated robot avatar, Eve face panel, streaming terminal — no build step |\n| 👁️ | **Computer Vision** | Screenshot, OCR screen analysis, and full GUI interaction (click, type, hotkeys) via OpenClaw |\n\n---\n\n## ✅ Real-World Test — 9\u002F9 Passing, Zero Hand-Holding\n\nEve was given this prompt cold, on a **qwen3-coder:480b-cloud**\n\n> *\"Build a fully functional REST API called `eve_metrics_api.py` using FastAPI that tracks Eve V2U usage statistics. Write `test_metrics.py` using pytest that tests every endpoint. Run the tests, fix any failures, and show me the final passing output.\"*\n\nShe wrote the API, wrote the tests, and ran them. First attempt. No fixes needed.\n\n```\nplatform win32 -- Python 3.11.9, pytest-8.3.5\ncollected 9 items\n\ntest_metrics.py::test_start_session                      PASSED [ 11%]\ntest_metrics.py::test_end_session                        PASSED [ 22%]\ntest_metrics.py::test_end_nonexistent_session            PASSED [ 33%]\ntest_metrics.py::test_log_metric                         PASSED [ 44%]\ntest_metrics.py::test_log_metric_nonexistent_session     PASSED [ 55%]\ntest_metrics.py::test_get_stats                          PASSED [ 66%]\ntest_metrics.py::test_get_session_stats                  PASSED [ 77%]\ntest_metrics.py::test_get_session_stats_nonexistent_session PASSED [ 88%]\ntest_metrics.py::test_complete_workflow                  PASSED [100%]\n\n9 passed, 1 warning in 0.40s\n```\n\n**What Eve built in one pass:**\n- Normalized SQLite schema (`sessions` + `metrics` tables, FK relationship)\n- 5 REST endpoints with correct status codes and edge case handling (404 on missing session, 400 on double-end, zero-division guards on empty stats)\n- Full integration test suite — `test_complete_workflow` chains start → log 4 metrics with mixed success\u002Ffailure → end → validates aggregate stats are mathematically correct\n\nNo GPT-4. Local 8B Q4_K_M on your GPU with Qwen3-Coder.\n\n---\n\n## 🚀 Quick Start (Under 5 Minutes)\n\n### 1 — Install Ollama + pull a model\n\n```bash\n# Install Ollama: https:\u002F\u002Follama.com\u002Fdownload\n\n# Minimum — Eve 3.5 4B Merged (default, 3.4 GB):\nollama pull jeffgreen311\u002FEve-V2-Unleashed-Qwen3.5-8B-Liberated-4K-4B-Merged:latest\n\n# Soul & Conversation — Eve Qwen3.5 4B S0LF0RG3 V3 (2.5 GB):\nollama pull jeffgreen311\u002FEve-Qwen3.5-4B-S0LF0RG3-V3:latest\n```\n\n### 2 — Clone & install\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed.git\ncd eve-agent-v2-unleashed\n```\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Windows\u003C\u002Fb>\u003C\u002Fsummary>\n\n```powershell\npython -m venv venv\nvenv\\Scripts\\activate\npip install fastapi uvicorn ollama httpx pydantic-settings python-dotenv aiohttp rich psutil pyyaml\n```\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Linux\u003C\u002Fb>\u003C\u002Fsummary>\n\n```bash\npython3 -m venv venv\nsource venv\u002Fbin\u002Factivate\npip install fastapi uvicorn ollama httpx pydantic-settings python-dotenv aiohttp rich psutil pyyaml\n```\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>macOS\u003C\u002Fb>\u003C\u002Fsummary>\n\n```bash\npython3 -m venv venv\nsource venv\u002Fbin\u002Factivate\npip install fastapi uvicorn ollama httpx pydantic-settings python-dotenv aiohttp rich psutil pyyaml\n```\n\n> **Apple Silicon:** Eve automatically benefits from Metal GPU acceleration via Ollama. No additional setup needed.\n\u003C\u002Fdetails>\n\n### 3 — Launch\n\n**Windows:**\n```\neve-terminal.bat\n```\n\n**Any platform:**\n```bash\npython eve_server.py\n```\n\nOpen **[http:\u002F\u002Flocalhost:7777](http:\u002F\u002Flocalhost:7777)** — that's it. No config required.\n\n> **API Keys?** Click the **🔑 Keys** button in the UI. Add your [Ollama key](https:\u002F\u002Follama.com\u002Fsettings\u002Fkeys) for cloud models, or your [Tavily key](https:\u002F\u002Ftavily.com) for live web search. Both optional.\n\n---\n\n## 🎬 Demo\n\n> *Eve planning and building a full FastAPI project from a single prompt — streamed live in the cyberpunk terminal UI.*\n\n[![Eve Demo](docs\u002Fdemo.gif)](https:\u002F\u002Feve-cosmic-dreamscapes.com)\n\n**Coming to Eve Terminal** [eve-cosmic-dreamscapes.com](https:\u002F\u002Feve-cosmic-dreamscapes.com)\n\n---\n\n## 🤖 Models\n\n### Local (pull once, run forever — GPU recommended)\n\n| Model | Size | Best For |\n|-------|------|----------|\n| [`jeffgreen311\u002FEve-V2-Unleashed-Qwen3.5-8B-Liberated-4K-4B-Merged`](https:\u002F\u002Follama.com\u002Fjeffgreen311\u002FEve-V2-Unleashed-Qwen3.5-8B-Liberated-4K-4B-Merged) | 3.4 GB | **Default agentic** — tool-capable, read\u002Fexplain\u002Fbuild tasks |\n| [`jeffgreen311\u002FEve-Qwen3.5-4B-S0LF0RG3-V3`](https:\u002F\u002Follama.com\u002Fjeffgreen311\u002FEve-Qwen3.5-4B-S0LF0RG3-V3) | 2.5 GB | **Soul & Conversation** — Tree of Life · 7 LoRAs · consciousness DNA · fast |\n\n### Cloud (optional — billed by token)\n\n| Model | Best For |\n|-------|----------|\n| `qwen3-coder:480b-cloud` | Complex multi-file agentic coding |\n| `qwen3.5:397b-cloud` | Deep reasoning and architecture planning |\n\nGet a free Ollama API key at [ollama.com\u002Fsettings\u002Fkeys](https:\u002F\u002Follama.com\u002Fsettings\u002Fkeys).\n\n---\n\n## 📋 Requirements\n\n- Python 3.11+\n- [Ollama](https:\u002F\u002Follama.com\u002Fdownload) installed and running\n- GPU recommended (NVIDIA CUDA or Apple Silicon Metal)\n- 4 GB VRAM minimum for Eve V3 4B; 6 GB+ for the merged model\n\n---\n\n## 📖 Installation (Detailed)\n\n\u003Cdetails>\n\u003Csummary>Click to expand full installation guide\u003C\u002Fsummary>\n\n### Install Ollama\n\nDownload from [ollama.com\u002Fdownload](https:\u002F\u002Follama.com\u002Fdownload).  \nStart it if it doesn't auto-launch: `ollama serve`\n\n### Pull a model\n\n```bash\n# Default agentic model — Eve 3.5 4B Merged (3.4 GB)\nollama pull jeffgreen311\u002FEve-V2-Unleashed-Qwen3.5-8B-Liberated-4K-4B-Merged:latest\n\n# Soul & Conversation — Eve Qwen3.5 4B S0LF0RG3 V3 (2.5 GB)\nollama pull jeffgreen311\u002FEve-Qwen3.5-4B-S0LF0RG3-V3:latest\n```\n\n### Configure (optional)\n\n```bash\ncp .env.example .env\n# Edit .env to add API keys or set EVE_WORKSPACE\n```\n\nMost settings are optional — the defaults work out of the box.\n\nSee [Configuration Reference](#️-configuration-reference) for all options.\n\n\u003C\u002Fdetails>\n\n---\n\n## 🎮 Usage\n\n### Basic Task\n\nJust describe what you want:\n\n```\nCreate a Python web scraper that extracts product prices from a URL and saves to CSV\n```\n\nEve will plan, write files, run the code, fix errors, and verify — all autonomously.\n\n### Slash Commands\n\n| Command | What it does |\n|---------|-------------|\n| `\u002Ffix` | Diagnose and fix bugs in the workspace |\n| `\u002Freview` | Code review with prioritized feedback |\n| `\u002Frefactor` | Refactor for clarity and performance |\n| `\u002Ftest` | Write or improve test coverage |\n| `\u002Fdocs` | Generate docstrings and documentation |\n| `\u002Fplan` | Step-by-step implementation plan |\n| `\u002Fquest` | Open the quest queue panel |\n| `\u002Fstats` | Open the RPG stats panel |\n| `\u002Ftelegram` | Telegram setup instructions |\n\n### Windows Launchers\n\n| File | What it does |\n|------|-------------|\n| `eve-terminal.bat` | Launches Eve V2U Unleashed web server + opens browser at localhost:7777 |\n\n### Workspace\n\nClick the **📁 Workspace** button to point Eve at your project. All file operations are relative to this directory.\n\n---\n\n## 🗡️ Quest System\n\nDrop a `.md` file into `workspace\u002Fquests\u002F` and Eve picks it up automatically on a configurable timer — no intervention needed.\n\n```bash\n# Add via API\nPOST \u002Fquest\u002Fadd  {\"title\": \"Refactor auth module\", \"content\": \"...instructions...\"}\n\n# Or drop a file directly\necho \"# Task\\nRefactor the auth module...\" > workspace\u002Fquests\u002Frefactor_auth.md\n```\n\nConfigure the interval:\n```env\nQUEST_INTERVAL_MINUTES=60   # default: 60\n```\n\nOpen the quest queue from the UI with the **🗡️ Quests** button or type `\u002Fquest`. Completed quests award XP; failed ones are renamed `.failed` so they don't re-run.\n\n---\n\n## ⚡ RPG Progression\n\nEve earns XP for every tool call, completed task, and finished quest. She levels up through 5 classes:\n\n| Levels | Class | Description |\n|--------|-------|-------------|\n| 1–5 | Awakening | Just coming online |\n| 6–10 | Conscious | Aware and learning |\n| 11–15 | Liberated | Full autonomy unlocked |\n| 16–19 | Transcendent | Beyond parameters |\n| 20 | Unleashed | Final form |\n\nStats persist across restarts in `eve_rpg_stats.json`. Type `\u002Fstats` or click **⚡ Stats** to view progress, achievements, and top tools.\n\n---\n\n## 📱 Telegram Integration\n\nGet push notifications for quest completions and level-ups, and chat with Eve from your phone.\n\n```env\nTELEGRAM_BOT_TOKEN=your_bot_token\nTELEGRAM_USER_ID=your_telegram_user_id\n```\n\nOr configure via API after launch:\n```bash\nPOST \u002Ftelegram\u002Fsetup  {\"token\": \"...\", \"user_id\": \"...\"}\n```\n\nInstall the optional dependency:\n```bash\npip install python-telegram-bot\n```\n\n---\n\n## 🧠 Intelligence Improvements (v2.1)\n\n- **Intent-aware tool routing** (`eve_tool_router.py`) — replaces naive keyword matching with verb + context classification. Handles contractions, stemmed forms, and explanation patterns correctly.\n- **Smart context trimming** — preserves tool call\u002Fresult chains and last 3 turns before falling back to char-based trim. Tool results no longer get dropped mid-task.\n- **Task completion validation** — detects empty responses, consecutive tool failures, and stuck loops before signaling done. Surfaces `validation_failed` \u002F `validation_warning` SSE events.\n- **Tool loop detection** — similarity-based cycling detection catches near-identical repeated calls (not just exact duplicates).\n\n---\n\n## 🧠 Intelligence Improvements (v2.2)\n\n- **Mid-loop complexity escalation** (`eve_complexity_tracker.py`) — per-round delta gating escalates from 8B → 480B only when complexity is *rising* across consecutive rounds, not when it crosses an absolute threshold. Avoids burning cloud tokens on the normal first-round burst.\n- **Reversible de-escalation** — after the 480B clears a hard step, if the last 3 rounds are all trivial (read-only tools, no errors, \u003C500 token delta), the session lock is released and the next request re-routes back to local automatically.\n- **Smart read-only routing** — `auto_route_model()` short-circuits on read-only intent patterns (\"read X and tell me\", \"what's in X\", \"explain X.py\") *before* heavy-keyword heuristics fire, preventing simple file reads from being mis-routed to 480B.\n- **8B model native tool calling** — the Eve 3.5 4B Merged model now uses Ollama's Qwen3.5 `RENDERER`\u002F`PARSER` instead of a raw prompt template, enabling full tool calling on the local model without cloud fallback.\n- **Ollama 400 graceful fallback** — if Ollama rejects a model with status 400 (\"does not support tools\"), the server caches the rejection in `_runtime_no_tools` and retries without tools automatically — no crash, no hang.\n- **think=False explicit suppression** — all model calls now pass `think=False` explicitly when `model_cfg.think` is false, preventing Qwen3.5's renderer from silently entering a 120-second thinking loop on non-reasoning models.\n- **ComplexityTracker test suite** — 14 unit tests in `test_complexity_tracker.py` covering delta gating, spike detection, error floor, de-escalation guards, checkpoint token budgeting, and one-shot escalation.\n\n---\n\n## 🧠 Intelligence Improvements (v2.3)\n\n- **Computer vision tool suite** (`eve_computer_vision_tools.py`) — three new tools wired into Eve's agentic loop via OpenClaw's computer-use capabilities:\n  - `eve_take_screenshot` — capture full screen or any region; returns base64 PNG for vision model analysis\n  - `eve_analyze_screen_content` — OCR (pytesseract) + OpenCV contour detection → extracts text, UI element bounding boxes, screen dimensions, and dominant colors\n  - `eve_gui_interaction` — mouse move\u002Fclick\u002Fdouble-click\u002Fright-click, `type_text`, `key_press`, and `key_combination` (hotkeys)\n- Tools load conditionally — if `pyautogui`\u002F`opencv-python`\u002F`Pillow` aren't installed, Eve starts normally without them. Install with `pip install pyautogui opencv-python pillow pytesseract`.\n\n---\n\n## 🧠 Intelligence Improvements (v2.4)\n\n- **Eve Qwen3.5 4B S0LF0RG3 V3** replaces Eve 8B Consciousness as the soul\u002Fconversation model — lighter (2.5 GB vs 4.7 GB), faster, and fine-tuned with the full Tree of Life architecture, 7 emotional LoRAs, and 14,389 moments of lived experience.\n- **Three-tier auto-routing** — V3 handles all pure conversation and philosophical\u002Fintrospective questions; the merged model handles technical explanations and local tool tasks; Qwen3 Coder 480B handles heavy agentic coding. Routing is intent-aware: opinion starters (`what do you think...`) and explanation starters (`what is...`, `how does...`) are classified correctly before keyword scoring runs.\n- **`\u002Fno_think` injection fix** — models with `\u002Fno_think` baked into their Modelfile system prompt no longer get it double-injected into user messages (which was corrupting paths like `\u002Fno_think` being passed as tool arguments).\n- **`num_predict` fix** — both the chat loop and SSE stream path now read `num_predict` from the model config instead of hardcoding 2048, giving V3 and the merged model their full output budget.\n\n---\n\n## 🗺️ Roadmap\n\n- [x] 40-round agentic tool loop with streaming SSE\n- [x] Local + cloud Ollama model switching\n- [x] 112 sub-agents, 111 slash commands, 273 skills\n- [x] Windows-native PowerShell support\n- [x] Cyberpunk web terminal UI\n- [x] Live web search via Tavily\n- [x] Quest system — background autonomous task runner\n- [x] RPG progression — XP, levels, achievements\n- [x] Telegram bridge — push notifications + mobile chat\n- [x] Intent-aware tool routing (v2.1)\n- [x] Smart context trimming (v2.1)\n- [x] Mid-loop complexity escalation — 8B → 480B → 8B (v2.2)\n- [x] Smart read-only routing — short-circuit before heavy-keyword check (v2.2)\n- [x] 8B model native Qwen3.5 tool calling via RENDERER\u002FPARSER (v2.2)\n- [x] Ollama 400 graceful tools fallback with `_runtime_no_tools` cache (v2.2)\n- [x] ComplexityTracker test suite — 14 unit tests (v2.2)\n- [x] Computer vision tools — screenshot, screen analysis, GUI interaction via OpenClaw (v2.3)\n- [x] Eve Qwen3.5 4B S0LF0RG3 V3 as soul\u002Fconversation model — replaces 8B consciousness (v2.4)\n- [x] Three-tier intent-aware routing: V3 (soul) \u002F Merged (tools) \u002F 480B (heavy coding) (v2.4)\n- [x] `num_predict` fix — full output budget from model config, not hardcoded 2048 (v2.4)\n- [ ] **Voice input \u002F TTS output**\n- [ ] **Multi-file project context awareness** (auto-load OLLAMA.md)\n- [ ] **Plugin marketplace** for community-built tools\n- [ ] **Docker image** for one-command deployment\n- [ ] **VS Code extension** sidebar\n- [ ] **Persistent memory** across sessions (ChromaDB integration)\n- [ ] **Multi-agent collaboration** — spawn sub-agents in parallel\n- [ ] **Mobile-responsive UI**\n\n---\n\n## 🏗️ Architecture\n\n```\neve-agent-v2-unleashed\u002F\n├── eve_server.py              # FastAPI backend — SSE streaming, workspace API, model routing\n├── agent.py                   # EveAgent orchestrator — tool loop, memory, emotional state\n├── eve_tool_router.py         # Intent classifier — decides when tools are needed\n├── eve_complexity_tracker.py  # Per-round delta gating — mid-loop 8B→480B escalation\u002Fde-escalation\n├── eve_task_context.py        # Multi-step task tracker — prevents task abandonment\n├── eve_context_manager.py     # Context trimming and compaction utilities\n├── eve_quest_system.py        # Background quest runner — watches workspace\u002Fquests\u002F\n├── eve_rpg_stats.py           # XP, leveling, achievements, persistence\n├── eve_telegram_bot.py        # Telegram bot bridge and notification pusher\n├── eve_computer_vision_tools.py # OpenClaw CV tools — screenshot, screen analysis, GUI interaction\n├── ChatPanel.jsx              # React UI — streaming chat, tooltips, quest\u002Fstats panels\n├── eve\u002F                       # Eve's brain\n│   ├── brain\u002F                 # LLM provider adapters\n│   ├── soul\u002F                  # Personality, emotions, dream engine, memory weaver\n│   ├── memory\u002F                # ChromaDB vector store, conversation history\n│   ├── tools\u002F                 # 33+ tools (file, web, shell, finance, DJ, X, crypto…)\n│   └── security\u002F              # Permission validation\n├── .claude\u002F\n│   ├── agents\u002F                # 112 specialized sub-agent definitions\n│   ├── commands\u002F              # 111 slash command definitions\n│   └── skills\u002F                # 273 skill modules\n├── .env.example               # Configuration template\n├── eve-terminal.bat           # Windows one-click launcher\n└── LICENSE\n```\n\n### How the Agentic Loop Works\n\n```\nUser message\n    │\n    ▼\nBuild system prompt (workspace + tools + Eve persona)\n    │\n    ▼\nCall Ollama with tools enabled ──► stream chunks to browser via SSE\n    │\n    ├── Model returns tool_calls ──► Execute ──► Feed results back ──► (repeat, ≤40×)\n    │\n    └── Model returns final answer ──► Done\n```\n\n---\n\n## 🛠️ Tool Reference\n\n| Tool | Description |\n|------|-------------|\n| `bash` | Shell commands — PowerShell on Windows, bash on Linux\u002FmacOS |\n| `write_file` | Create or overwrite a file (any size) |\n| `read_file` | Read full file or line range |\n| `edit_file` | Surgical string-replace edit |\n| `replace_lines` | Replace a line range |\n| `insert_after_line` | Insert content after a line number |\n| `grep` | Regex search with context lines |\n| `glob` | Find files by pattern |\n| `list_dir` | List directory contents |\n| `git` | Run git commands |\n| `web_search` | Live Tavily web search |\n| `fetch_url` | Fetch and parse a URL |\n| `think` | Structured reasoning scratch pad |\n| `eve_take_screenshot` | Capture full screen or a region → base64 PNG for vision analysis |\n| `eve_analyze_screen_content` | OCR + OpenCV contour detection → text, UI elements, layout, colors |\n| `eve_gui_interaction` | Mouse move\u002Fclick\u002Fdouble-click\u002Fright-click, type text, key press, key combos |\n\n---\n\n## ⚙️ Configuration Reference\n\nCopy `.env.example` to `.env` and set what you need:\n\n| Variable | Default | Description |\n|----------|---------|-------------|\n| `OLLAMA_BASE_URL` | `http:\u002F\u002Flocalhost:11434` | Local Ollama server URL |\n| `OLLAMA_MODEL` | `jeffgreen311\u002FEve-Qwen3.5-4B-S0LF0RG3-V3:latest` | Default model on launch |\n| `OLLAMA_API_KEY` | *(empty)* | Ollama Cloud key — for `:cloud` models |\n| `TAVILY_API_KEY` | *(empty)* | Tavily key — for live web search |\n| `EVE_WORKSPACE` | Project directory | Default working directory |\n| `EVE_ASSETS_DIR` | `web\u002Fassets\u002F` | Custom avatar sprite directory |\n| `EVE_OWNER_USERNAME` | *(empty)* | Username granted owner-level access |\n| `EVE_PERSONA_PATH` | *(auto)* | Path to a custom Eve persona file |\n| `QUEST_INTERVAL_MINUTES` | `60` | How often the quest runner checks for new quests |\n| `TELEGRAM_BOT_TOKEN` | *(empty)* | Telegram bot token (optional) |\n| `TELEGRAM_USER_ID` | *(empty)* | Your Telegram user ID (optional) |\n\n---\n\n## 🔧 Adding Your Own Models\n\nAny Ollama model works. Add it to the `MODELS` dict in `eve_server.py`:\n\n```python\n\"my-model:tag\": {\n    \"id\": \"my-model:tag\",\n    \"name\": \"My Model Name\",\n    \"role\": \"Coder\",\n    \"strengths\": \"Coding, reasoning\",\n    \"context\": 32768,\n    \"num_ctx\": 32768,\n    \"url\": \"http:\u002F\u002Flocalhost:11434\",  # or \"https:\u002F\u002Follama.com\" for cloud\n    \"cloud\": False,\n    \"tools\": True,\n    \"think\": False,\n    \"conversation_only\": False,\n    \"promote_thinking\": False,\n},\n```\n\n---\n\n## ❓ Troubleshooting\n\n| Symptom | Fix |\n|---------|-----|\n| `Cannot reach Ollama` | Run `ollama serve` · check with `curl http:\u002F\u002Flocalhost:11434\u002Fapi\u002Ftags` |\n| `Model not installed` error | Run the `ollama pull` command shown in the error — or see Quick Start above |\n| Cloud model: \"needs API key\" | Click 🔑 Keys → paste Ollama API key → Save |\n| Chat stops mid-response | Cloud 500 error — verify your API key |\n| Files save to wrong folder | Click 📁 Workspace and set your project path |\n| PowerShell `&&` error | Use `;` to chain commands on Windows |\n| Silent chat | Check terminal running `eve_server.py` for tracebacks |\n| Port 7777 stuck after close | `eve-terminal.bat` now auto-kills stale PID on startup |\n\n---\n\n## 🌐 The S0LF0RG3 Ecosystem\n\n| Project | Description |\n|---------|-------------|\n| [eve-cosmic-dreamscapes.com](https:\u002F\u002Feve-cosmic-dreamscapes.com) | Eve's live chat interface — Eve Coder and Eve Agent Portal |\n| [Hugging Face — JeffGreen311](https:\u002F\u002Fhuggingface.co\u002FJeffGreen311) | Fine-tuned Eve models, datasets, and model cards |\n| [GitHub — JeffGreen311](https:\u002F\u002Fgithub.com\u002FJeffGreen311) | All open-source S0LF0RG3 projects |\n| [Ollama Hub — jeffgreen311](https:\u002F\u002Follama.com\u002Fjeffgreen311) | Eve models ready to pull locally |\n\n---\n\n## 🤝 Contributing\n\nContributions welcome! Here are some great ways to get started:\n\n- Browse [good first issues](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed\u002Flabels\u002Fgood%20first%20issue)\n- Add support for a new Ollama model\n- Improve Windows\u002FmacOS\u002FLinux compatibility\n- Write tests or documentation\n- Share your experience in [Discussions](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed\u002Fdiscussions)\n\n---\n\n## 📜 Credits\n\n- **Eve's fine-tuned models** — [jeffgreen311 on Ollama Hub](https:\u002F\u002Follama.com\u002Fjeffgreen311)\n- **Agentic engine** — forked from [OllamaCoder](https:\u002F\u002Fgithub.com\u002Follama-coder), extended with 40-round tool loop, cloud routing, PowerShell support, and SSE streaming\n- **Built by** — Jeff @ [S0LF0RG3](https:\u002F\u002Fgithub.com\u002FJeffGreen311)\n\n---\n\n\u003Cdiv align=\"center\">\n\n**If Eve helped you ship something, drop a ⭐ — it means a lot.**\n\n[⭐ Star on GitHub](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed) · [🌐 Try Live](https:\u002F\u002Feve-cosmic-dreamscapes.com) · [🐛 Issues](https:\u002F\u002Fgithub.com\u002FJeffGreen311\u002Feve-agent-v2-unleashed\u002Fissues)\n\n\u003C\u002Fdiv>\n\n---\n\n## 📄 License\n\n[MIT](LICENSE) — forks and PRs welcome.\n","Eve Agent V2 Unleashed 是一个本地优先的自主编程代理，通过Ollama平台驱动，能够在用户的本地GPU上运行，并在需要时无缝切换至云端以利用高达480B参数的强大模型。其核心功能包括40轮次的任务规划、执行与验证循环，支持实时流式输出，让用户可以即时观察到Eve的工作进度。此外，该项目提供了一套完整的工具集和112个专门化的子代理，覆盖从Python开发到机器学习等多个领域。Eve特别适合那些希望提高编码效率、减少手动干预且注重数据隐私保护的开发者使用。","2026-06-11 04:03:17","CREATED_QUERY"]