[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-76272":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":34,"readmeContent":35,"aiSummary":36,"trendingCount":16,"starSnapshotCount":16,"syncStatus":37,"lastSyncTime":38,"discoverSource":39},76272,"neuralinverse","NeuralInverse\u002Fneuralinverse","NeuralInverse","Code Modern. Code Legacy. Code Firmware. - open-source AI-native IDE with agentic coding, Power Mode, legacy modernization, and firmware development","https:\u002F\u002Fneuralinverse.com",null,"TypeScript",165,7,16,14,0,1,13,96,9,2.71,"Apache License 2.0",false,"main",true,[27,28,29,30,31,32,33],"agentic-coding","ai","code-editor","ide","llm","open-source","vscode-fork","2026-06-12 02:03:41","# NeuralInverse Community Edition\n\n\u003Cdiv align=\"center\">\n\t\u003Cimg\n\t\tsrc=\".\u002Fsrc\u002Fvs\u002Fworkbench\u002Fbrowser\u002Fparts\u002Feditor\u002Fmedia\u002Fneuralinverse_logo.png\"\n\t\talt=\"NeuralInverse\"\n\t\twidth=\"300\"\n\t\theight=\"300\"\n\t\u002F>\n\u003C\u002Fdiv>\n\n**Code Modern. Code Legacy. Code Firmware.**\n\nNeuralInverse CE is a free, open-source AI-native IDE built for the work most AI tools ignore - modernizing legacy systems, developing firmware, and migrating regulated codebases. Bring your own LLM, no cloud lock-in.\n\n- 🌐 [Website](https:\u002F\u002Fneuralinverse.com)\n- 📧 [Contact](mailto:github@neuralinverse.com)\n- 🏢 [Enterprise Edition](https:\u002F\u002Fneuralinverse.com) - compliance, GRC, and legacy modernization for regulated industries\n- [![Sponsor NeuralInverse](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FSponsor-NeuralInverse-ea4aaa?logo=github-sponsors&logoColor=white)](https:\u002F\u002Fgithub.com\u002Fsponsors\u002FNeuralInverse)\n\n\n## Install\n\n**macOS \u002F Linux**\n```bash\ncurl -fsSL neuralinverse.com\u002Fsh | bash\n```\n\n**Windows** (PowerShell)\n```powershell\nirm neuralinverse.com\u002Fwin | iex\n```\n\n\n## Features\n\n### AI Chat & Agentic Coding\n- Inline and sidebar chat with full codebase context\n- **Power Mode**: autonomous multi-step agent that plans, edits, runs commands, and iterates without hand-holding\n- **Bring Your Own LLM**: direct integration with Claude, GPT-4, Gemini, Ollama, Bedrock, and more - your API keys stay local, no proxy\n\n### Firmware & Embedded Development\nPurpose-built for embedded engineers - not just syntax highlighting:\n- **MCU database**: 357 MCU variants across 11 manufacturers (STM32, NXP, ESP32, Nordic, Renesas, TI, Infineon, RP2040, AURIX, and more) with full specs, memory maps, and peripheral lists\n- **SVD auto-load**: fetches CMSIS-SVD register maps on demand, parses peripheral trees with bit-field detail, handles `derivedFrom` inheritance\n- **Datasheet intelligence**: drop in a PDF and the agent extracts register maps, timing constraints, errata, and pinouts - 3-tier pipeline with LLM reclassification, content-hash caching (zero re-processing on re-open), and page-level citations\n- **22 `fw_*` agent tools**: register map queries, peripheral config, bit-field lookup, errata checks, timing constraints, build + flash, serial monitor, MISRA\u002FCERT-C checks, GDB debug (breakpoints, step, memory read\u002Fwrite), binary size analysis, code generation for peripheral init \u002F ISR \u002F DMA \u002F clock config\n- **Platform knowledge packs**: STM32, ESP32, nRF (Zephyr), RP2040 - clock trees, DMA patterns, ISR constraints, toolchain quirks injected into every agent session\n- **Build system support**: PlatformIO, CMake, ESP-IDF, Make, Zephyr, Arduino, Mbed, STM32CubeIDE - build and flash directly from the IDE\n- **Serial monitor**: Web Serial API, 10,000-line ring buffer, baud auto-detection, debug probe auto-detection (ST-Link, J-Link, CMSIS-DAP, FTDI, CH340, and more)\n- **Compliance-aware**: MISRA C:2012, CERT C, IEC 62304, ISO 26262 rules injected into the agent context so generated code respects safety constraints\n\n### Legacy Modernization\nA full 5-stage migration platform built into the IDE. Open it with `Cmd+Alt+M`.\n\n**Stage 1 - Discovery**\nScans source trees across 30+ languages and build systems - COBOL, PL\u002FSQL, RPG, Natural, FORTRAN, Assembler, AUTOSAR ARXML, CAN DBC, IEC 61131-3, TTCN-3, and all mainstream languages. Extracts:\n- Dependency graphs, cyclomatic complexity, tech debt categories (17 generic + 14 firmware\u002Findustrial)\n- Regulated data patterns: PCI-DSS, GDPR, HIPAA, SOX, financial-core, tax-compliance, ISO 26262, IEC 61850 GOOSE, 3GPP key material, CAN signal IDs\n- GRC snapshot: violation counts by domain and severity, blocking violations, top rule violations\n- Market-vertical detection: automotive (AUTOSAR\u002FMISRA), safety (IEC 61508\u002F61511), telecom (3GPP\u002FGSMA), energy (IEC 61850\u002FDNP3), industrial OT (IEC 62443), embedded MCU\n\n**Stage 2 - Planning**\nCPM critical path scheduling across 7 phases (foundation -> schema -> core-logic -> API layer -> integration -> compliance -> cutover):\n- 12+ blocker types including AUTOSAR RTE dependency, E2E protection gap, ASIL decomposition break, GOOSE protection relay, SIS\u002FSIL downgrade\n- Market-vertical compliance gates: automotive ASIL-D formal verification, energy GOOSE path isolation, telecom 3GPP key externalisation, IIoT\u002FOT IEC 62443 zone\u002Fconduit isolation\n- Effort estimation with safety-critical language surcharges (embedded-C, AUTOSAR, assembler, IEC 61131)\n- Stage 3 locked behind plan approval gate\n\n**Stage 3 - Source Resolution**\nPrepares each migration unit for translation by inlining dependencies:\n- COBOL copybook inliner with cycle detection, CALL graph resolver\n- PL\u002FSQL %TYPE\u002F%ROWTYPE inliner, Java @EJB\u002F@Autowired context injection\n- RPG \u002FCOPY+\u002FINCLUDE expansion, Natural USING DA\u002FCALLNAT resolution\n- Generic import inliner for TypeScript, Python, Go, Rust, C#, Kotlin, Scala and more\n- Leaf-node-first scheduling with risk priority and concurrency control\n\n**Stage 4 - Translation**\nLanguage-pair profiles with deep idiom mappings:\n- 36 profiles including: COBOL->Java (32 idioms), PL\u002FSQL->TypeScript, RPG->Java, Natural->Java, FORTRAN->Python, Angular 1->18, Vue 2->3\n- 25 firmware profiles: bare-metal C->FreeRTOS, bare-metal C->Zephyr, AUTOSAR CP->AP, CAN DBC->CANopen, IEC 61850->OPC-UA, O-RAN C->Go, EtherCAT->EtherCAT NG, TTCN-3->pytest, and more\n- 6-priority context budget (source, type\u002Fnaming, interfaces, rule patterns, rules\u002Fglossary, annotations)\n- Verification checks: non-empty, no placeholders, no truncation, balanced braces, length sanity\n- Decision tracking: every IRaisedDecision (naming, type, rule interpretation) stored and reviewable\n\n**Knowledge Base**\nPersistent store surviving IDE restarts - every translation decision, type mapping, glossary term, annotation, compliance gate result, checkpoint, and audit log entry. Importable\u002Fexportable.\n\n**Stage 5 - Cutover**\n- 8-point readiness gate (4 blocking, 2 warning, 2 info)\n- Audit bundle export with FNV-1a chain integrity hash and verifiable bundle integrity\n- Committed file writing to target tree via VS Code file service\n\n**Autonomy Engine**\nOptional autonomous execution: auto-approval policies, concurrent sub-task scheduling, batch progress events, configurable iteration limits and command blocklists via `.neuralinverseagent` config.\n\n### Multi-model\nSwitch models per task - use a fast model for chat, a powerful one for agentic runs.\n\n\n## What is not in CE\n\nThe following features are available in [NeuralInverse Enterprise](https:\u002F\u002Fneuralinverse.com):\n\n- neuralInverseChecks - real-time GRC and compliance enforcement (HIPAA, SOC2, FDA 21 CFR Part 11, ISO 26262, etc.)\n- Checks Agent - AI agent with programmatic access to violations, rule explanations, and compliance reporting\n- NeuralInverse auth and team collaboration features\n\n\n## Credits\n\nNeuralInverse CE is built on top of [VS Code](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fvscode) by Microsoft. We are grateful to the VS Code team and all upstream contributors.\n\n\n## Architecture\n\nNeuralInverse CE is built on [VS Code](https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fvscode).\n\nKey modules:\n- `src\u002Fvs\u002Fworkbench\u002Fcontrib\u002Fvoid\u002F` - AI agent and chat infrastructure\n- `src\u002Fvs\u002Fworkbench\u002Fcontrib\u002FpowerMode\u002F` - Power Mode agentic workflows\n- `src\u002Fvs\u002Fworkbench\u002Fcontrib\u002FneuralInverseModernisation\u002F` - Legacy code modernization platform\n- `src\u002Fvs\u002Fworkbench\u002Fcontrib\u002FneuralInverseFirmware\u002F` - Firmware datasheet knowledge base\n\n\n## Building from source\n\n```bash\nnpm install\nnpm run compile\n```\n\nSee [HOW_TO_CONTRIBUTE.md](.\u002FHOW_TO_CONTRIBUTE.md) for full setup instructions including platform prerequisites and developer mode.\n\n\n## License\n\nCopyright 2026 Neural Inverse Inc. Licensed under the Apache License 2.0. See [License.txt](.\u002FLicense.txt) for details.\n\n\n## Support\n\n- Email: github@neuralinverse.com\n- Website: https:\u002F\u002Fneuralinverse.com\n","NeuralInverse 是一个开源的AI原生IDE，专为现代化遗留系统、开发固件及迁移受监管代码库而设计。其核心功能包括通过自带LLM实现的AI聊天和自主编码，支持多种微控制器和嵌入式开发工具，并具有强大的数据表智能解析能力。特别适用于需要对老旧系统进行升级或在严格监管环境下工作的场景，如医疗设备、汽车电子等行业。此外，它还提供了丰富的构建系统支持和串行监控工具，帮助开发者更高效地完成从代码编写到调试的全过程。",2,"2026-06-11 03:54:52","CREATED_QUERY"]