{"slug": "hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices", "title": "HashCortX – AI workspace app, Now works on pre-2013 CPU windows devices", "summary": "HashCortX v2.5.0, a local-first AI workspace app, now runs on pre-2013 CPU Windows devices, according to its developer. The MIT-licensed desktop app supports 11 cloud AI providers plus Ollama, includes an autonomous coding agent and multi-agent swarms, and has no backend, telemetry, or account requirements. The latest release, dated 17 August 2026, is a 43 MB DMG for Apple Silicon, with 138 commits since v2.0.0.", "body_md": "**A local-first AI workspace — multi-provider chat, an autonomous coding agent, and multi-agent swarms in one native desktop app.**\n\n**No backend. No telemetry. No account. MIT-licensed.**\n\nEight workspaces — chat, an autonomous coding agent, multi-agent swarms, financial document analysis, a security scanner, a business-app builder, 3D planning and a virtual project desktop — behind one window, with nine specialist agents and a real Python sandbox working inside them.\n\nEvery AI request goes straight from your machine to the provider whose key you entered. Nothing passes through HashCortx infrastructure, because there is no HashCortx infrastructure. Point it at Ollama and it runs with the network off — everything except the Python sandbox, which fetches its runtime on first use.\n\nType |\nNative desktop app (Tauri v2) |\nRuns on |\nmacOS Apple Silicon — built and used daily. Windows — tested on Windows 10, with an installer in the release. Linux compiles and passes its tests in CI, but nobody has run the app there yet |\nLicense |\nMIT |\nLatest release |\nv2.5.0 (17 August 2026) — 43 MB DMG for Apple Silicon, 80 MB installed. 33 MB of that is the bundled embedding model, and most of the rest is the runtime that executes it |\nAI providers |\n11 cloud (Groq, Gemini, OpenAI, Anthropic, Moonshot, DeepSeek, Mistral, Cerebras, SambaNova, OpenRouter, NVIDIA NIM) + Ollama |\nStack |\nRust · vanilla JavaScript · no bundler · no framework · ~35,700 lines JS, ~4,530 Rust |\nTests |\n93 Rust tests, run by CI on Linux, macOS and Windows · 1,384 source checks, every one of them run by CI on every push |\nTelemetry · backend · accounts |\nNone · None · None |\n\nv2.5.0 is what this page describes.It carries 138 commits since v2.0.0 — the offline knowledge base, the security work, Windows and Linux support, and a long list of features that were advertised and did not run.[What changed], including what is still open.\n\n**Nothing phones home.** No analytics, no crash reporting, no update pings. The only outbound connections are to providers you configured yourself.\n\n**Your keys, your models.** Eleven cloud providers and Ollama, configured at once, switched freely, mixed inside a single swarm run.\n\n**The agent asks before it acts.** File and shell calls hit a Rust permission gate and a compiled denylist that no prompt can talk its way past.\n\n**Search that understands meaning.** Ask about \"stopping a runaway command\" and your notes about killing a process on timeout come back — from a model that ships inside the app and never sends anything anywhere.\n\n**You can audit it.** MIT, no build step, no minified application code. Read it, fork it, ship your own.\n\n| Workspace | What it does | |\n|---|---|---|\n| 01 | Chats |\nMulti-provider chat with projects, attachments, slash commands, full history |\n| 02 | Coder |\nThe coding agent: file tree, real file edits, shell access, browser panel |\n| 03 | 3D Forge |\nDescribe a part, get a dimensioned solid you can print. Fuses to one watertight body, cuts real holes, exports in millimetres |\n| 04 | Finance |\nStatements, CSV, PDF and XLSX into KPIs and charts. Never invents a number |\n| 05 | Sandbox |\nAgents scanning untrusted code for malware, prompt injection, suspicious logic |\n| 06 | ERP |\nDescribe a workflow, get a working interactive prototype |\n| 07 | Agent Swarm |\nChain mode, vote mode, automatic provider failover mid-run |\n| 08 | Virtual OS |\nA simulated project desktop an agent works inside |\n\nTwo more surfaces are not workspaces and have no tab of their own: **Agents** — the nine built-in specialists and the builder for your own — opens as a menu over the message you are writing, and **Split** — one prompt, two models, streamed side by side — is a toggle inside chat.\n\nThe nine agents, the Python sandbox and every workspace in detail: [MODES_GUIDE.txt](/Hash-7777/HashCortX/blob/main/MODES_GUIDE.txt) · [Wiki → Features](https://github.com/Hash-7777/HashCortX/wiki/Features)\n\nThe agent reads your real files, edits them, runs commands, and shows every change as a diff you can expand. It does not get to do any of that quietly.\n\nEvery filesystem and shell call passes through `HC.guard.request()`\n\nand lands in Rust, where a compiled denylist refuses anything touching `~/.ssh`\n\n, `~/.aws`\n\n, `~/.gnupg`\n\n, the system directories, or HashCortx's own stored keys — whether the path arrives as a file operation or inside a shell command. Inside the folder you opened, the agent works without interrupting you. Outside it, everything asks first, **including reads**, because an agent that reads a file is an agent that can send it to a provider.\n\nEvery command is bounded: a five-minute timeout, closed stdin, a 512 KB output cap. Full detail, and the honest limits: [SECURITY.md](/Hash-7777/HashCortX/blob/main/docs/SECURITY.md).\n\nChain mode hands each agent's output to the next. Vote mode runs one prompt across several models and has a judge score the answers. If a provider rate-limits or dies mid-run, the swarm swaps to another one you configured and carries on with the same context.\n\nAnything you ingest becomes searchable by **meaning**, not just by matching words — `bge-small-en-v1.5`\n\n(MIT) ships inside the app and runs natively in Rust. It is inference-only: a sentence encoder, not a language model. On an x86-64 machine without AVX2 the app is built without it and searches by keyword instead; see [Older processors](#older-processors-without-avx2).\n\n**Nothing is fetched.** No first-run download, no cache to warm. It works offline on first launch.**Nothing is sent.** What you index never crosses a network boundary.**34 MB of the download.** That is the price of the two lines above, paid once.\n\nResults are ranked by meaning and by keyword at once, then fused — so a rare error code still finds its exact match while a paraphrased question still finds the right passage.\n\nDownload the DMG from the [latest release](https://github.com/Hash-7777/HashCortX/releases/latest), open it, drag HashCortx to `/Applications`\n\n.\n\nThe DMG is built for **Apple Silicon**. On an Intel Mac, build from source.\n\nOn **Windows** there is an installer in the [latest release](https://github.com/Hash-7777/HashCortX/releases/latest). It is the build made without local embeddings, so it starts on a processor of any age and searches the knowledge base by keyword rather than by meaning; for search by meaning on an AVX2 processor, build with the default features (below). On **Linux** there is no prebuilt download — build from source (below). Check your processor first: the default build needs AVX2 and BMI2, and there is a build that does not. See [Older processors](#older-processors-without-avx2).\n\nThe build is unsigned and not notarised, so on first launch right-click the app and choose **Open**, then **Open** again. If macOS still refuses:\n\n```\nxattr -dr com.apple.quarantine /Applications/HashCortx.app\n```\n\nThen open **Settings → API keys**, add a key, press **Test**. Or skip keys entirely and run a model on your own machine — **Settings → Local model** walks you through it and checks each step for you.\n\n```\ncd ~                   # start in your own folder, not wherever the shell opened\ngit clone https://github.com/Hash-7777/HashCortX.git\ncd HashCortX\nnpm install\nnpm run tauri dev      # live-reload development\nnpm run tauri build    # DMG in src-tauri/target/release/bundle/dmg/\n```\n\nNode 18+ and a Rust toolchain via `rustup`\n\n, plus **macOS**: Xcode Command Line Tools · **Linux**: Ubuntu 24.04+ and the Tauri v2 system libraries (glibc 2.38+ is required to link the bundled ONNX Runtime) · **Windows**: MSVC build tools and WebView2.\n\n**On Windows, run cd ~ before cloning — the line above is not decoration.**\nPowerShell opened as an administrator starts in\n\n`C:\\Windows\\System32`\n\n, so a\npasted `git clone`\n\nlands inside Windows' own system folder. The prerequisite\ninstallers want an administrator window; the build itself does not, and a normal\nPowerShell already starts somewhere sensible.A checkout under `System32`\n\nfails in a way that points nowhere near the cause.\nThe WiX tools Tauri bundles with are 32-bit, and a 32-bit process reading that\npath is redirected by Windows to `SysWOW64`\n\n, where the checkout does not exist —\nso the Rust build succeeds and bundling then fails saying it cannot find a file\nthat is plainly there. The folder also inherits System32's permissions, so\ngetting rid of it afterwards needs an administrator:\n\n```\nrobocopy C:\\Windows\\System32\\HashCortX $HOME\\HashCortX /E /XD target\nRemove-Item -LiteralPath C:\\Windows\\System32\\HashCortX -Recurse -Force\n```\n\nThe default build links a prebuilt ONNX Runtime for the embedding model. It is\ncompiled for x86-64 processors with **AVX2 and BMI2** — Intel Haswell (2013),\nAMD Excavator (2015) and newer — and it is linked statically, so its start-up\ncode runs before `main()`\n\n. On an older processor it executes an instruction the\nCPU does not have and the process is killed while it is still loading: no\nwindow, no error, nothing on screen.\n\nBuild without it, and the app starts on any x86-64 machine:\n\n```\nnpx tauri build -- --no-default-features\n```\n\nThe `--`\n\nmatters: the Tauri CLI has no such flag of its own and passes\neverything after it to cargo. Through an npm script it takes two, because npm\neats the first one — `npm run tauri build -- -- --no-default-features`\n\n.\n\nThat build has no embedding model in it. The knowledge base still works and\nstill searches, by keyword rather than by meaning; `embed_available`\n\nreports\nfalse so the interface can say so rather than quietly returning worse results.\nThe binary is also far smaller — about 20 MB against about 80 MB, measured on\nApple Silicon — since neither the model nor the runtime is compiled in. Apple\nSilicon is unaffected either way.\n\nBefore pushing, run what CI runs:\n\n```\nnpm run check                                     # 2,968 checks over the real source\ncargo test --manifest-path src-tauri/Cargo.toml   # 97 tests\n```\n\n| Layer | Technology |\n|---|---|\nShell |\nTauri v2 — Rust core, the system webview, no Chromium |\nBackend |\nRust: filesystem, shell, audit log, usage log, embeddings, Keychain migration |\nSecurity |\nCompiled denylist in `security/denylist.rs` , permission prompt via `HC.guard.request()` |\nFrontend |\nVanilla JavaScript. No React, no TypeScript, no bundler, no build step |\nEmbeddings |\nbge-small-en-v1.5 (MIT) compiled into the binary, run via ONNX Runtime |\nPython |\nPyodide (CPython on WebAssembly) with pandas, numpy, matplotlib, python-docx, openpyxl, reportlab |\nVendored libs |\nmarked, highlight.js, DOMPurify, mermaid, pdf.js, jsPDF, three.js — all local, no CDN |\n\nNo bundler is a deliberate constraint. The interface is about 2.5 MB of source that ships as written, so any reader can follow a feature from the button that triggers it to the Rust function that performs it, without a source map. What makes the download large is the embedding model and the runtime that executes it, not the app.\n\n[ARCHITECTURE.md](/Hash-7777/HashCortX/blob/main/docs/ARCHITECTURE.md) · [SECURITY.md](/Hash-7777/HashCortX/blob/main/docs/SECURITY.md) · [CONTRIBUTING.md](/Hash-7777/HashCortX/blob/main/CONTRIBUTING.md) · [CHANGELOG.md](/Hash-7777/HashCortX/blob/main/CHANGELOG.md)\n\n**No backend, no telemetry, no accounts, no auto-updater.** The binary makes no network call except to the provider endpoints you set up.\n\n**A permission gate in Rust.** Sensitive paths are denied unconditionally, whether they arrive as a file operation or inside a shell command. Every guarded action is logged to `~/.hashcortx/audit.log`\n\n.\n\n**Keys are not encrypted.** They sit in an app-scoped local directory protected by your user account, not by Keychain encryption — because a Keychain item's access list is bound to the code signature, and an unsigned build would re-prompt for every key on every update. Code signing is on the roadmap; the reasoning is written out in full in [SECURITY.md](/Hash-7777/HashCortX/blob/main/docs/SECURITY.md).\n\n**Measured usage, not guessed.** One JSON line per response to `~/.hashcortx/usage.jsonl`\n\n— timestamp, model id, token counts. No prompt, no answer, no file names. Counts come from the provider's own metadata; if a provider reports none, HashCortx writes nothing rather than estimating. [HashMeterAi](https://github.com/Hash-7777/HashMeterAi) reads that file if you install it.\n\n```\njq -s 'map(.input_tokens + .output_tokens) | add' ~/.hashcortx/usage.jsonl\n```\n\n**Source-grounded modes.** Published Papers Researcher, Medical Lexi-Check and Finance are constrained never to fabricate data.\n\nBest effort as of August 2026. If something is out of date, [open an issue](https://github.com/Hash-7777/HashCortX/issues/new/choose).\n\n| HashCortx | Cursor | Claude Code | Continue | Aider | Cline | Zed | |\n|---|---|---|---|---|---|---|---|\n| Type | Native app | VS Code fork | CLI | Extension | Terminal CLI | Extension | Native editor |\n| License | MIT | Proprietary | Proprietary | Apache 2.0 | Apache 2.0 | Apache 2.0 | GPL/AGPL |\n| Free | Bring your own key | Subscription | Subscription or API | Yes | Yes | Yes | Yes |\n| Cloud providers | 11 | Limited | Anthropic only | Many | Many | Many | Several |\n| Local models (Ollama) | Yes | Limited | No | Yes | Yes | Yes | Yes |\n| Multi-agent swarms | Yes | No | No | No | No | No | No |\n| Workspaces beyond coding | 8 | No | No | No | No | No | No |\n| Built-in specialist agents | 9 | None | None | None | None | None | None |\n| Telemetry | None | Yes | Opt-out | Opt-in | None | None | Opt-in |\n\n**Is it free?** Yes. MIT, no paid tier, no usage caps. You pay the AI providers directly, or nothing at all with Ollama.\n\n**Does it work offline?** Yes, with Ollama. The knowledge base works offline regardless. Cloud providers need the internet.\n\n**Which systems?** macOS Apple Silicon is built and used daily. **Windows runs — tested on Windows 10**, where the app has been built, installed and used; that release carries a Windows installer. Linux compiles and passes its tests in CI on every push, but nobody has launched it there, so treat Linux as buildable rather than supported.\n\n**What processor does it need?** On x86-64 — every Windows and Linux machine, and Intel Macs — the default build needs **AVX2 and BMI2**: Intel Haswell (2013) or AMD Excavator (2015) and newer. That is not the app itself but the ONNX Runtime it links to run the embedding model, which is compiled for those instructions and starts up before any of the app's own code. On an older processor the app cannot start at all, and because the failure happens during loading there is no window and no message — a double-click that appears to do nothing. Building with `--no-default-features`\n\nremoves that runtime and produces an app that starts on any x86-64 machine; see [Older processors](#older-processors-without-avx2). Apple Silicon is unaffected.\n\n**Does it send my code anywhere?** Only to the provider you configured, when you send a message. There is no HashCortx server.\n\n**Are my API keys encrypted?** No — see above.\n\n**Was it built with AI?** Yes, heavily. Roughly 30 million tokens across Claude, GPT and other frontier models during the v2.0.0 build, and substantially more over the 138 commits since, under human architecture, review and correction. Disclosed because HashCortx is itself an AI tool, and hiding that would be incoherent. Every product decision — the workspace structure, the local-first rule, the Permission Guard, the swarm failover pattern, the source-grounding constraints — is the author's.\n\nMore at [Wiki → FAQ](https://github.com/Hash-7777/HashCortX/wiki/FAQ).\n\n- A release cut from\n`main`\n\n, so the download matches this README - Code signing and notarisation, which also unlocks Keychain key storage\n- Someone actually running the app on Linux and Windows — compiling and passing tests is not the same thing\n- Continued extraction of\n`app.js`\n\ninto focused modules - Permission Guard coverage for Virtual OS and 3D Forge\n- Reaching the knowledge base from Coder mode, which still cannot see it\n\nSuggest something in [Issues](https://github.com/Hash-7777/HashCortX/issues/new/choose) or [Discussions](https://github.com/Hash-7777/HashCortX/discussions).\n\nFour local-first apps, same principles — no cloud, no telemetry, your data stays where it is.\n\n| App | What it is | Licence |\n|---|---|---|\nHashCortx (you are here) |\nThe local-first AI workspace | MIT |\nHashCerebrum |\n\n**HashMeterAi****HashNotch** They interlock, through files on your disk rather than a service:\n\n- HashCortx appends real token counts to\n`~/.hashcortx/usage.jsonl`\n\n, and**HashMeterAi** reads it — so your spend across every tool is measured in one place, by software that never phones home. - When a run finishes, HashCortx posts a short notice to\n`~/.hashnotch/activities.json`\n\n, and**HashNotch** lights up the notch — a title and nothing else, never a model name, a prompt or an answer. If it is not installed, the file simply sits there unread.\n\n`Cmd/Ctrl + Shift + C`\n\ntoggle Coder · `Cmd/Ctrl + Shift + N`\n\nnew chat · `Cmd/Ctrl + K`\n\nmodel picker\n\n**Author** — [Seif Hashish](https://github.com/Hash-7777), independent open-source developer with a pharma and clinical background, which is where the refusal-to-fabricate constraints in the medical and finance modes come from. · [hashcortx.com](https://hashcortx.com)\n\n**License** — MIT. See [LICENSE](/Hash-7777/HashCortX/blob/main/LICENSE).\n\n**HashCortx** · One window · Twelve providers · Zero data leak · Local-first · MIT\n\n[Download](https://github.com/Hash-7777/HashCortX/releases/latest) · [Wiki](https://github.com/Hash-7777/HashCortX/wiki) · [Discussions](https://github.com/Hash-7777/HashCortX/discussions)", "url": "https://wpnews.pro/news/hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices", "canonical_source": "https://github.com/Hash-7777/HashCortX", "published_at": "2026-08-31 23:43:24+00:00", "updated_at": "2026-08-31 23:52:00.301857+00:00", "lang": "en", "topics": ["ai-products", "ai-agents", "ai-tools", "developer-tools"], "entities": ["HashCortX", "Groq", "Gemini", "OpenAI", "Anthropic", "Moonshot", "DeepSeek", "Mistral"], "alternates": {"html": "https://wpnews.pro/news/hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices", "markdown": "https://wpnews.pro/news/hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices.md", "text": "https://wpnews.pro/news/hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices.txt", "jsonld": "https://wpnews.pro/news/hashcortx-ai-workspace-app-now-works-on-pre-2013-cpu-windows-devices.jsonld"}}