byollm 0.1.0-alpha.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Of Tomorrow, Inc.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,118 @@
+# `byollm`
+
+What end users run. Connects **outbound** to an app you trust, claims only the
+jobs you have agreed to run, and executes them on your own models.
+
+```bash
+npx byollm@alpha connect https://your-app.com
+```
+
+There is nothing to open on your network. The daemon never listens.
+
+## Five minutes, start to finish
+
+You need a model server. Ollama is the usual one:
+
+```bash
+ollama serve            # http://127.0.0.1:11434
+ollama pull gemma3:12b
+```
+
+Then:
+
+```bash
+npx byollm@alpha connect https://your-app.com
+```
+
+```
+  Open:  https://your-app.com/settings/runners
+  Code:  KRTZ-9F2Q      (expires in 10m)
+
+  waiting for approval… ✓ paired as you@example.com
+```
+
+You approve inside the app's own login session — the daemon never asks for a
+password and never accepts a pasted secret.
+
+## Configuration
+
+`~/.byollm/config.json`. Everything the daemon will ever do is in this file.
+
+```jsonc
+{
+  "backends": {
+    // One HTTP backend covers Ollama, MLX, llama.cpp and vLLM — they all
+    // speak OpenAI-compatible /v1/chat/completions.
+    "local": {
+      "backend": "openai-http",
+      "baseUrl": "http://127.0.0.1:11434/v1",
+    },
+    "mlx": { "backend": "openai-http", "baseUrl": "http://127.0.0.1:8080/v1" },
+    "claude": { "backend": "claude-cli" },
+  },
+  "routes": {
+    "llm.generate": { "backend": "local", "model": "gemma3:12b" },
+    "llm.chat": { "backend": "claude", "model": "claude-opus-5" },
+  },
+  "concurrency": 2,
+}
+```
+
+A job's `kind` selects a route **you defined**. A job can never name a model, a
+URL, a path or a flag — there is no field on the wire for any of them.
+
+`byollm backends` shows what is configured, what is healthy, and what is
+therefore advertised. A backend that is down is never advertised, so you never
+get work you cannot run.
+
+## The trust surface
+
+The meter is the product, and it gets the same care as the loop.
+
+```bash
+byollm status         # what's connected, what's running, what you've done for others
+byollm log            # every prompt that has ever run here
+byollm log --full     # the whole text, not the first line
+byollm pause          # stop claiming work
+byollm resume
+```
+
+Every prompt is appended to `~/.byollm/ingress.log` **before** it executes, so
+a job that wedges the machine still leaves a record of what it was. The file is
+JSONL, `0600`, and yours to read, grep and delete.
+
+## Lending your machine to other people
+
+Off by default. A fresh daemon runs your work and nobody else's.
+
+```bash
+byollm allow https://your-app.com alice   # asks you to confirm, in plain words
+byollm allow --list                       # everyone who can use this machine
+byollm disallow https://your-app.com alice
+```
+
+The allowlist is **yours**, checked locally, keyed by `(app, user)`. An app
+saying "this runner is allowed" is not enough — your daemon decides. Community
+jobs are additionally rate-limited, capped daily, given a tighter resource
+budget, and their prompts are reduced to hashes after 7 days so you are not
+holding strangers' content indefinitely.
+
+**Your subscription-backed models are never part of this.** `claude-cli` is
+locked to your own work — a protocol rule, not a setting you can change.
+
+## Security
+
+Every payload is treated as hostile input. Breakout is structurally
+impossible: process-class backends get a fixed argv with the prompt on stdin,
+a stripped environment, an empty scratch directory and hard timeout/output
+caps; HTTP-class backends spawn nothing at all. The model has no tools, no
+retrieval and no MCP.
+
+Prompt injection — steering what the model _says_ — is not prevented by
+anything and we do not claim otherwise. It is bounded here because the model
+has no tools and the output is inert.
+
+Full threat model, including what the OS stops us dropping:
+[`docs/security.md`](../../docs/security.md).
+
+MIT

package/dist/bin.d.ts

@@ -0,0 +1 @@
+#!/usr/bin/env node

package/dist/bin.js

@@ -0,0 +1,8 @@
+#!/usr/bin/env node
+import {
+  main
+} from "./chunk-S5WHDSUF.js";
+
+// src/bin.ts
+process.exitCode = await main(process.argv.slice(2));
+//# sourceMappingURL=bin.js.map

package/dist/bin.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/bin.ts"],"sourcesContent":["#!/usr/bin/env node\n/**\n * The `byollm` executable.\n *\n * Deliberately the only module in this package that *does* something on\n * import. `cli.ts` stays a pure module so it can be imported as a library —\n * and so the bundler cannot hoist the command implementations into a shared\n * chunk and leave this entry with nothing to run, which is exactly what\n * happened when the two were the same file.\n */\nimport { main } from \"./cli.js\";\n\nprocess.exitCode = await main(process.argv.slice(2));\n"],"mappings":";;;;;;AAYA,QAAQ,WAAW,MAAM,KAAK,QAAQ,KAAK,MAAM,CAAC,CAAC;","names":[]}