![\"Pipe\"](\"website/logo.svg\"){kind=logo}
\n Open the Pipe Playground\n \n
Pipe — The MCP-native runtime, production-ready\n\n[](https://github.com/MachuraHarry/pipe/actions/workflows/ci.yml)\n[](LICENSE)\n[](https://github.com/MachuraHarry/pipe/releases)\n[](#)\n[](#model-context-protocol)\n[](https://github.com/mcp/MachuraHarry/pipe)\n[](https://registry.modelcontextprotocol.io/?q=MachuraHarry)\n\n> **The first language with built-in MCP — server and client. 246 builtins, single ~8 MB binary. Zero dependencies.**\n> **Officially listed in the [official MCP Registry](https://registry.modelcontextprotocol.io/?q=MachuraHarry)** (v1.1.1, active). One-click install from [GitHub MCP Registry](https://github.com/mcp/MachuraHarry/pipe) for Copilot & VS Code.\n\n## What's New in v1.0\n\nPipe v1.0.0 is the **production-ready release**, consolidating the entire v0.9.x series:\n\n- **Guard clauses** — `| pattern if cond -> body` in match expressions\n- **Concurrency primitives** — channels (`send`/`recv`/`try_recv`), mutex (`lock`/`unlock`), counting semaphore (`acquire`/`release`)\n- **Bytecode-VM improvements** — constant folding, alias import namespaces, bytecode cache\n- **MQTT 5.0 module** — pure Pipe MQTT client with input validation, CONNACK properties, DISCONNECT handling\n- **docs-pipe** — RAG module for documentation-native search with heading-aware chunking\n- **Test framework** — setup/teardown hooks, `assert_near`/`assert_contains`, VM test blocks\n- **Hardened sandbox** — audit rounds 1-6, deterministic env masking, central egress gate\n- **246 builtins** — 36 AI + 13 MCP + 192 standard, up from 226 in v0.9.3\n- **23 modules** — MQTT, SQLite, pipe-http, pipe-web, pipe-orm, pipe-cli, and more\n\n## Quick Install\n\n```sh\ncurl -fsSL https://pipe-lang.com/install.sh | bash # Linux & macOS\n```\n\nWindows (PowerShell): `irm https://pipe-lang.com/install.ps1 | iex`\n\nThe installer downloads the latest release, verifies its SHA256 checksum and installs `pipe` into `~/.local/bin` (or `/usr/local/bin` when run as root). Pin a version with `PIPE_VERSION=v1.0.0`. See the [full install docs](docs/en/01-getting-started.md).\n\nInstalled copies update themselves against the latest GitHub release: `pipe --update` (or `pipe --update-check` to only look, `pipe --version` to show what you are on). The updater verifies the release checksum and replaces the binary in place.\n\n## Privacy & DSGVO\n\nPipe is **DSGVO-konform / GDPR-compliant by design**:\n\n- **Zero telemetry & analytics** — the binary never phones home, nothing leaves your machine\n- **Self-hosted single binary** — runs entirely on your infrastructure\n- **No cloud** — no vendor server processes your data\n- **Open source (MIT)** — fully auditable\n- **Local AI** — with Ollama, not a single byte leaves your network; cloud providers are used only if you configure one\n\n## The Problem\n\nRunning AI in production is harder than it should be:\n\n- **Security** — LLMs with file access, network, and `exec` are a liability. You need fine-grained sandboxing at the language level, not afterthought middleware.\n- **Performance** — Sequential API calls turn a 1-second pipeline into a 10-second bottleneck. Parallelism shouldn't require `asyncio.gather()` boilerplate.\n- **Vendor Lock-in** — Switching from OpenAI to DeepSeek means rewriting your Python SDK code. Provider changes should be one line.\n- **Tool Integration** — Connecting LLMs to external tools (GitHub, databases, filesystems) is a maze of SDKs and API wrappers. MCP should be a language primitive, not a library.\n\n**Pipe fixes this at the language level.**\n\n## What is Pipe?\n\nPipe is a **Semantic Pipeline Runtime (SPR)** — a pipeline-native language where `summarize`, `translate`, and `classify` sit on the same syntax level as `+`, `sort`, and `len`. Data flows top to bottom through composable transformations. One binary. Zero dependencies.\n\n**Python + LangChain (~80 lines):**\n\n```python\nimport openai\nclient = openai.OpenAI()\ndef summarize(text):\n r = client.chat.completions.create(model=\"gpt-4o\", messages=[{\"role\":\"user\",\"content\":text}])\n return r.choices[0].message.content\ndef translate(text, lang):\n r = client.chat.completions.create(model=\"gpt-4o\",\n messages=[{\"role\":\"system\",\"content\":f\"Translate to {lang}\"},{\"role\":\"user\",\"content\":text}])\n return r.choices[0].message.content\ntext = open(\"news.txt\").read()\nprint(translate(summarize(text), \"de\"))\n```\n\n**Pipe (5 lines):**\n\n```pipe\nread_file \"news.txt\"\n > summarize -- LLM call\n > translate \"de\" -- LLM call\n > print\n```\n\n## Model Context Protocol\n\nPipe has **built-in MCP** — both as a server and client. No SDKs, no npm packages, no Python. Pure Go stdlib.\n\n### MCP Server — Expose your tools\n\n```pipe\nfn get_weather city\n match city\n | \"Berlin\" -> \"22°C, sunny\"\n | \"London\" -> \"15°C, rainy\"\n | _ -> city ++ \": no data\"\n\nai_tool \"get_weather\" \"Get weather for a city\" {city: \"City name\"} get_weather\nmcp_server \"Weather Agent\" \"1.0.0\"\nmcp_serve_stdio\n```\n\nConfigure in Claude Desktop (`claude_desktop_config.json`):\n\n```json\n{ \"mcpServers\": { \"pipe\": { \"command\": \"/tmp/pipe\", \"args\": [\"agent.pipe\"] } } }\n```\n\n### MCP Client — Use external tools\n\n```pipe\nai_provider \"deepseek\"\nai_set_key \"deepseek\" (env \"DEEPSEEK_API_KEY\")\n\n-- Connect to GitHub + Filesystem MCP servers\nmcp_use_stdio \"npx\" \"-y\" \"@modelcontextprotocol/server-github\" {GITHUB_TOKEN: (env \"GITHUB_TOKEN\")}\nmcp_use_stdio \"npx\" \"-y\" \"@modelcontextprotocol/server-filesystem\" \"/tmp\"\n\n-- AI discovers and uses all tools automatically\nresult: ai_with_tools \"You are a DevOps assistant.\" \"Search pipe's open issues and list files in /tmp.\" 10\nprint result\n```\n\n**Any stdio MCP server** works immediately: Filesystem, GitHub, Git, Postgres, SQLite, Slack, Brave Search, Memory, Sequential Thinking — anything on npm/uvx.\n\n## Use Cases\n\n### Log Analysis → Incident Report\n\n```pipe\nis_critical: fn line\n contains line \"critical\"\n\nread_file \"/var/log/app/errors.log\"\n > split \"\\n\"\n > filter is_critical\n > summarize\n > translate \"de\"\n > save \"incident_report.txt\"\n```\n\n### RAG Pipeline\n\n```pipe\nai_provider \"deepseek\"\n\ndocs: read_lines \"knowledge_base.txt\"\nvectors: embed_batch docs\n\nquestion: \"How does the bytecode VM work?\"\nq_vec: embed question\ntop: nearest q_vec vectors 3\n\ncontext: \"\"\nfor idx in top\n context: context ++ (at docs idx) ++ \"\\n---\\n\"\n\nask (\"Context:\\n\" ++ context ++ \"\\nQuestion: \" ++ question)\n > print\n```\n\n### AI Agent with Tool Calling\n\n```pipe\nfn get_weather city\n match city\n | \"Berlin\" -> \"22°C, sunny\"\n | \"London\" -> \"15°C, rainy\"\n | _ -> city ++ \": no data\"\n\nai_tool \"get_weather\" \"Get current weather for a city\" {city: \"Name of the city\"} get_weather\n\nai_with_tools \"You are a weather assistant.\" \"What's the weather in Berlin and London?\"\n > print\n```\n\n### Concurrency — 3 LLM Calls in 1.5s, Not 4s\n\n```pipe\nai_provider \"deepseek\"\n\na: \"Explain monads\" >> ask\nb: \"What is CP/M?\" >> ask\nc: \"Explain RFC 791\" >> ask\n\nprint a ++ b ++ c -- Future auto-resolution\n```\n\n### Discord CI/CD Notifications\n\n```pipe\nimport \"discord.pipe\" as d\nai_provider \"deepseek\"\n\n-- AI code review per commit, sent as Discord embed\nreview: ai_chat \"Review this code change\" diff 800\n\nd.d_webhook_embed (env \"DISCORD_WEBHOOK\") {\n title: \"CI: Push to master\",\n color: 3447003,\n fields: [\n {name: \"Changed Files\", value: stat},\n {name: \"AI Review\", value: review}\n ]\n}\n```\n\n## Comparison: Pipe vs Python + LangChain\n\n| | Python + LangChain | Pipe |\n|--------------------------|-------------------------------|--------------------------------|\n| **RAG pipeline** | ~80 LOC | ~8 LOC |\n| **Sandbox LLM access** | Custom middleware | One `sandbox_profile` block |\n| **Switch AI provider** | Rewrite SDK calls | `ai_provider \"deepseek\"` |\n| **Deploy to server** | Docker + venv + pip | `scp pipe binary` |\n| **Parallel LLM calls** | `asyncio.gather()` boilerplate | `>>` operator, `ai_batch` |\n| **MCP Server + Client** | Library-dependent | 13 builtins, zero deps, 100+ servers |\n| **Binary size** | ~500 MB (with deps) | ~8 MB |\n\n## Features\n\n- **MCP-native** — 13 builtins for MCP Server + Client. Pure Go stdlib. Connect to any stdio MCP server\n- **Ship AI pipelines 10x faster** — 36 AI + 13 MCP builtins: no imports, no SDKs, no API wrappers\n- **Lock down AI agents in one line** — Declarative sandbox profiles: restrict `exec`, `write_file`, `http_get` with a single block\n- **Deploy in seconds** — One statically-linked ~8 MB binary. No venv, no pip, no Docker. Linux, macOS, Windows, Raspberry Pi, or your browser via WebAssembly\n- **3 LLM calls in 1.5s, not 4s** — `>>` starts any pipeline stage in the background. Futures auto-resolve. `ai_batch` handles hundreds of texts concurrently with built-in rate limiting\n- **No vendor lock-in** — OpenAI, Anthropic (Claude), DeepSeek, Ollama. Switch with one line. Same code works everywhere\n- **Concurrency primitives** — channels (`send`/`recv`), mutex (`lock`/`unlock`), counting semaphore (`acquire`/`release`)\n- **Pipeline-native syntax** — `>` sequential, `>>` parallel. Data flows top to bottom — readable, composable, debuggable\n- **Social platforms built in** — Discord webhooks and Telegram bots as Pipe modules. AI code reviews, notifications, chat — zero API costs for sending\n- **Bytecode VM** — Compile to bytecode, run on a stack VM with automatic caching. Measured 0.6x-55x vs tree-walker depending on workload (recursion-heavy code up to ~55x)\n- **Module ecosystem** — 23 curated modules, registry with version pinning (`@1.0.0`). `pipe -get` installs, import by name\n- **Built-in testing** — `test` blocks with `assert_eq`, `assert_error`. Run with `pipe -test`. Zero setup\n- **GitHub Action** — Run Pipe directly in CI/CD. No installation needed\n- **VSCode Extension** — Syntax highlighting, IntelliSense, LSP-powered diagnostics and completions\n- **Self-extracting binary** — Ship your pipeline as a standalone executable (`pipe -build`)\n\n## Quick Start\n\n```bash\ngit clone https://github.com/MachuraHarry/pipe && cd pipe && make build\nexport DEEPSEEK_API_KEY=\"sk-...\"\n./bin/pipe -vm -q -c 'ai_provider \"deepseek\"; ask \"What makes Pipe different?\" > print'\n```\n\n## Try it in your browser\n\nNo install needed — Pipe runs fully in your browser via WebAssembly:\n\n\n\n```pipe\n-- Paste this into the playground and hit Run\nlevels: [\"error\",\"warn\",\"info\"]\nread_file \"server.log\"\n > classify levels\n > summarize\n > print\n```\n\n## GitHub Action\n\nRun Pipe directly in CI/CD — no installation needed:\n\n```yaml\n- uses: MachuraHarry/pipe/.github/actions/pipe-action@master\n with:\n script: |\n print \"Hello from CI/CD!\"\n log: exec \"git log --oneline -20\"\n print (get log \"output\")\n```\n\n[→ GitHub Action Documentation](docs/en/20-github-action.md)\n\n## VSCode Extension\n\nSyntax highlighting and full IntelliSense for `.pipe` files, powered by a Language Server Protocol client (`vscode/`) and the `pipe-lsp` server (`cmd/pipe-lsp`):\n\n- Completion, hover docs, signature help, go-to-definition, references, rename\n- Diagnostics (parse errors, undefined/unused variables) and semantic highlighting\n- Format document, auto-completion of brackets, auto-indent and code folding\n\n```sh\nmake vsix # builds the server and packages vscode/pipe-syntax-1.0.0.vsix\n```\n\nOr run the extension in development with F5 from the `vscode/` folder. See [VSCode Extension Documentation](docs/en/15-vscode-extension.md).\n\n## Module Ecosystem\n\nPipe has a [curated module library](https://github.com/MachuraHarry/pipe-modules) — **23 reusable modules** with version pinning:\n\n| Infrastructure | Data & CLI | AI & Agents | DevTools | Social |\n|---|---|---|---|---|\n| `pipe-http` | `sqlite` | `rag-pipe` | `pipe-test` | `telegram-bot` |\n| `pipe-cli` | `jpipe` | `log-analyzer` | `pipe-validate` | `mqtt` |\n| `pipe-orm` | `pipe-tpl` | `sentiment` | | |\n| `pipe-web` | `pipe-date` | `code-review` | | |\n| | | `translate-batch` | | |\n| | | `changelog-gen` | | |\n| | | `email-classifier` | | |\n| | | `incident-report` | | |\n| | | `parallel-runner` | | |\n| | | `date-formatter` | | |\n| | | `docs-pipe` | | |\n\n```bash\npipe -search # Browse modules\npipe -search sql # Filter by keyword\npipe -get sqlite # Install latest\npipe -get sqlite@0.8.0 # Install specific version\n```\n\n```pipe\nimport \"sqlite\" -- database engine\nimport \"pipe-http\" -- HTTP client\nimport \"mqtt\" -- MQTT 5.0 client\nimport \"discord.pipe\" as d -- Discord webhooks + bot\n\nidx: index_create h \"knowledge\"\nindex_add idx \"Pipe is an AI-native language.\"\nindex_search idx \"language\" 3 > each print\n```\n\n[→ Ecosystem Documentation](docs/en/21-ecosystem.md) | [→ Contribute a Module](https://github.com/MachuraHarry/pipe-modules/blob/master/CONTRIBUTING.md)\n\n## Execution Modes\n\n| Mode | Command | Speed |\n|------|---------|-------|\n| Tree-Walker | `./bin/pipe script.pipe` | Baseline |\n| Bytecode VM | `./bin/pipe -vm -q script.pipe` | 0.6x-55x (recursion-heavy up to ~55x) |\n\n## 49 AI + MCP Builtins (36 AI + 13 MCP)\n\n### Understanding\n`summarize`, `translate`, `classify`, `extract`, `ask`, `generate`, `generate_json`\n\n### Speed & Control\n`ai_stream`, `ai_batch`, `ai_parallel`, `ai_rate_limit`, `ai_chat`, `ai_chat_json`\n\n### Search & Retrieval\n`web_search`, `wiki_search`, `embed`, `embed_batch`, `cosine_sim`, `dot_product`, `nearest`\n\n### Agents & Tools\n`agent`, `agent_ask`, `agent_clear`, `ai_tool`, `ai_with_tools`\n\n### Config & Cost\n`ai_provider`, `ai_model`, `ai_host`, `ai_set_key`, `ai_timeout`, `ai_cache`, `ai_cost`, `ai_tokens`, `ai_cache_hits`, `ai_cache_misses`\n\n### MCP — Model Context Protocol\n`mcp_server`, `mcp_serve_stdio`, `mcp_serve_sse`, `mcp_tools`, `mcp_resource`, `mcp_resource_template`, `mcp_prompt`, `mcp_resources`, `mcp_read_resource`, `mcp_prompts`, `mcp_prompt_get`, `mcp_use_stdio`, `mcp_use_sse`\n\n### Self-Healing\n`try_ai`, `try_ai_log`\n\n## Advanced Features\n\n### Self-Healing Code (`try_ai`)\n```pipe\nai_provider \"deepseek\"\n\nresult: try_ai\n \"42\" * 3 -- E002 Type Error -> AI wraps with to_num -> 126\ncatch e\n 0 -- only reached if AI fix fails\n\nprint result -- 126\n```\n\n### Parallel Pipeline (`>>`)\n```pipe\na: \"Frage A\"\n >> ask\nb: \"Frage B\"\n >> ask\nc: \"Frage C\"\n >> ask\n\nprint a ++ b ++ c -- Future auto-resolution\n```\n\n### Guard Clauses in Match\n```pipe\nfn classify severity\n match severity\n | s if s > 9 -> \"critical\"\n | s if s > 5 -> \"warning\"\n | _ -> \"info\"\n```\n\n### Concurrency: Channels\n```pipe\nch: chan 3\ngo { send ch \"hello\" }\ngo { send ch \"world\" }\nprint (recv ch) ++ \" \" ++ (recv ch)\n```\n\n### Sandbox Profiles\n```pipe\nsandbox_profile \"safe\" {fs: \"read-only\", network: false, exec: false, ai: true}\nsandbox_profile \"agent\" {fs: \"temp-only\", network: true, exec: false, ai: true}\n\nset_sandbox \"safe\"\nread_file \"/etc/config\" -- reading allowed\nwrite_file \"/etc/config\" -- E_SANDBOX blocked\n```\n\n## Architecture\n\n```\nSource (.pipe) -> Lexer -> Parser -> AST -> [ Tree-Walker | Compiler + VM ]\n |\n Builtins (246 total: 36 AI + 13 MCP + 192 standard)\n |\n MCP Server <-> MCP Clients (stdio + HTTP)\n```\n\n- 67 token types, 36 AST node types, 43 opcodes\n- ~37,000 LoC Go, 643 tests, 87 example programs\n- Zero dependencies — pure Go stdlib\n\n## Documentation\n\n[→ Full documentation (English)](/docs/en/index.md)\n[→ Vollständige Dokumentation (Deutsch)](/docs/de/index.md)\n\n## Project Structure\n\n```\npipe/\n├── cmd/\n│ ├── pipe/main.go # Entry point\n│ └── pipe-lsp/ # Language Server Protocol server (IntelliSense)\n├── pkg/\n│ ├── ai/ # AI provider integrations\n│ ├── analysis/ # IntelliSense library (builtins, diagnostics, completion...)\n│ ├── ast/ # AST node definitions\n│ ├── build/ # Self-extracting binary builder\n│ ├── cache/ # Bytecode cache\n│ ├── compiler/ # Compiler to bytecode\n│ ├── eval/ # Tree-walk interpreter\n│ ├── formatter/ # Code formatter\n│ ├── gen/ # Code generation helpers\n│ ├── lexer/ # Lexer and tokens\n│ ├── mcp/ # MCP server + client (zero-dependency)\n│ ├── object/ # Runtime objects\n│ ├── parser/ # Parser\n│ ├── stdlib/ # Standard library helpers\n│ └── vm/ # Bytecode VM\n├── examples/ # 87 example programs\n├── test/integration/ # Integration tests\n├── vscode/ # VSCode extension (syntax highlighting + LSP client)\n├── docs/ # Documentation (DE + EN)\n├── website/ # Project website\n├── modules/ # Language modules (mqtt, discord, x, etc.)\n├── Makefile\n├── go.mod\n└── LICENSE\n```\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md).\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n",
"bytes": 18202,
"sha": "23e5fefc27c7d285692e4c90dcde533e53faad076eb8af0853e807d92bef4ae0",
"repo_slug": "machuraharry/pipe",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/mcp_io_github_machuraharry_pipe_docs_39a44a8c/readme"
}