brasskey 0.1.0a1__tar.gz

brasskey-0.1.0a1/.gitignore

@@ -0,0 +1,27 @@
+# Python
+__pycache__/
+*.py[cod]
+*.egg-info/
+build/
+dist/
+
+# Environments (uv creates .venv; uv.lock IS committed)
+.venv/
+
+# Tool caches and reports
+.pytest_cache/
+.ruff_cache/
+.coverage
+.coverage.*
+coverage.xml
+htmlcov/
+
+# Editors and OS
+.idea/
+.vscode/
+*.swp
+.DS_Store
+
+# Local secrets (API keys come from env vars or the OS keyring; never commit them)
+.env
+.env.*

brasskey-0.1.0a1/CONTRIBUTING.md

@@ -0,0 +1,98 @@
+# Contributing to Brasskey
+
+Thanks for your interest! Brasskey is pre-alpha and moving fast toward v0.1 "It runs". This page
+covers how to set up, the rules every change follows, and how contributions are licensed.
+
+## Scope first
+
+We build one milestone at a time (see the Status section of the README). Anything outside the
+current milestone waits; something moves earlier only if a design partner needs it and something of
+equal size moves later. If you want to propose a bigger change, open an issue or write a short
+proposal in `docs/decisions/` before writing code.
+
+## Set up
+
+You need [uv](https://docs.astral.sh/uv/) 0.13.6 or newer (`uv self update`).
+
+```bash
+uv sync                      # create .venv with runtime + dev dependencies from uv.lock
+uv run brasskey --help       # run the CLI from source
+```
+
+uv uses its own managed Python builds for this project (`python-preference = "only-managed"`), so
+an active conda or system Python does not leak into the environment.
+
+## Everyday commands
+
+```bash
+uv run pytest                                   # unit tests; no GPU needed
+uv run pytest -m gpu                            # real NVIDIA tests (opt-in)
+uv run pytest -m engine                         # real engine + small model download (opt-in)
+uv run ruff check . && uv run ruff format --check .
+uv run pyright                                  # strict for src/ and scripts/
+uv run python scripts/check_licenses.py         # dependency license policy
+uv build --build-constraints scripts/build-constraints.txt --require-hashes --no-create-gitignore
+python3 scripts/check_dist.py dist              # what a release would publish
+```
+
+CI runs all of these on Linux and macOS (Apple Silicon) with Python 3.11 to 3.14. Releases are
+published by CI only; see [RELEASING.md](RELEASING.md).
+
+## Rules every change follows
+
+- **Tests:** every feature ships with tests that pass on a laptop without a GPU. Use fake hardware
+  probes and fake engine processes; mark real-hardware tests `@pytest.mark.gpu` (or `engine`,
+  `network`).
+- **License header:** every `.py` file starts with `# SPDX-License-Identifier: Apache-2.0` (a test
+  checks this).
+- **Engines stay out of process:** never import vLLM, SGLang, MLX, llama.cpp bindings, PyTorch or
+  GuideLLM into Brasskey. They run as subprocesses or containers behind an adapter. Ruff enforces
+  this (`TID251`).
+- **CLI contract:** every command supports `--json`, ends with a one-line `Next:` suggestion,
+  explains errors in plain language with a fix, and shows tracebacks only with `--debug`.
+- **Measure, don't guess:** recommendations are backed by a measurement or clearly labeled as an
+  estimate.
+- **Privacy:** never log or store prompt content, API keys or tokens.
+- **Secure defaults:** servers bind to `127.0.0.1`; any other address requires an API key. Keys
+  come from environment variables or the OS keyring and never appear on command lines, in logs or
+  in lock files. Telemetry is off; nothing leaves the machine without explicit opt-in.
+- **Reproducible deployments:** `brasskey.yaml` (intent) and `brasskey.lock` (resolved plan and
+  measured results) are Pydantic v2 models with exported JSON Schemas. After changing them, run
+  `uv run python scripts/export_schemas.py` and commit `schemas/`.
+- **Supply chain:** GitHub Actions are pinned by full commit SHA and tools by exact version, and
+  `uv.lock` is committed (see also Dependencies below).
+- **Naming:** "Brasskey" in prose, `brasskey` in code and commands. No `bk` alias.
+
+## Dependencies
+
+Ask before adding a dependency (open an issue). It must be Apache-2.0, MIT or BSD compatible;
+GPL and AGPL code or dependencies are never accepted, and code is never copied from GPL or AGPL
+projects. Record the dependency and its license in `NOTICE`. The lock only resolves releases that
+have been public for at least 7 days (`exclude-newer` in `pyproject.toml`).
+
+## Decisions
+
+Record notable decisions as short ADRs in `docs/decisions/NNNN-title.md` using the template there.
+
+## Pull requests
+
+Keep pull requests small and focused, fill in the template checklist, and make sure CI is green.
+Write commit messages in the imperative mood ("Add hardware probe for Apple Silicon").
+
+## Contributor License Agreement
+
+Brasskey uses a Contributor License Agreement (CLA) based on the Apache Software Foundation's
+individual and corporate CLAs: you keep the copyright in your contribution and grant Prcept AI
+LLP a broad license (including patents) to use and distribute it. A CLA bot will ask you to sign
+once, on your first pull request.
+
+The CLA bot is not set up yet, so **we cannot merge outside pull requests until it is**. Issues,
+bug reports and discussions are very welcome in the meantime.
+
+**Our pledge:** the Brasskey core is licensed under Apache 2.0 and stays Apache 2.0. Paid
+features for organizations live in separate repositories and never change the license of this one.
+
+## Conduct and security
+
+Everyone taking part follows our [Code of Conduct](CODE_OF_CONDUCT.md). Report security problems
+privately as described in [SECURITY.md](SECURITY.md), never in a public issue.

brasskey-0.1.0a1/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

brasskey-0.1.0a1/NOTICE

@@ -0,0 +1,65 @@
+Brasskey
+Copyright 2026 Prcept AI LLP
+
+This product includes software developed at Prcept AI LLP
+(https://github.com/Prcept-AI/brasskey), licensed under the Apache License, Version 2.0.
+
+-------------------------------------------------------------------------------
+Inference engines and tools (launched as separate processes; not bundled)
+-------------------------------------------------------------------------------
+
+Brasskey starts the following projects as separate processes or containers. They are not
+included in, linked into or redistributed with Brasskey, and they keep their own licenses:
+
+  llama.cpp          MIT          https://github.com/ggml-org/llama.cpp
+  MLX                MIT          https://github.com/ml-explore/mlx
+  mlx-lm             MIT          https://github.com/ml-explore/mlx-lm
+  vLLM               Apache-2.0   https://github.com/vllm-project/vllm
+  SGLang             Apache-2.0   https://github.com/sgl-project/sglang
+  GuideLLM           Apache-2.0   https://github.com/vllm-project/guidellm
+
+On macOS, mlx-lm is run through uv (MIT OR Apache-2.0, https://github.com/astral-sh/uv) in an
+isolated, pinned environment when it isn't already installed; GuideLLM is run the same way
+(with CPU-only PyTorch, BSD-3-Clause). uv is not bundled either.
+
+Container images for vLLM and SGLang may contain NVIDIA CUDA libraries under NVIDIA's own
+license terms. Users pull those images themselves; Brasskey does not redistribute them.
+
+-------------------------------------------------------------------------------
+Python dependencies (installed separately by pip or uv; not bundled)
+-------------------------------------------------------------------------------
+
+  typer              MIT          https://github.com/fastapi/typer
+                     (includes a vendored copy of Click, BSD-3-Clause)
+  rich               MIT          https://github.com/Textualize/rich
+  pydantic           MIT          https://github.com/pydantic/pydantic
+  psutil             BSD-3-Clause https://github.com/giampaolo/psutil
+  nvidia-ml-py       BSD-3-Clause https://pypi.org/project/nvidia-ml-py/
+                     (NVIDIA's Python bindings for NVML; the NVML library itself ships
+                     with the user's NVIDIA driver and is not redistributed)
+  httpx2             BSD-3-Clause https://github.com/pydantic/httpx2
+                     (with httpcore2, BSD-3-Clause, and truststore, MIT)
+  PyYAML             MIT          https://github.com/yaml/pyyaml
+  platformdirs       MIT          https://github.com/tox-dev/platformdirs
+  huggingface_hub    Apache-2.0   https://github.com/huggingface/huggingface_hub
+                     (with hf-xet, Apache-2.0; tqdm, MPL-2.0 AND MIT, used unmodified)
+  uvicorn            BSD-3-Clause https://github.com/Kludex/uvicorn
+                     (with h11, MIT, and click, BSD-3-Clause)
+  Starlette          BSD-3-Clause https://github.com/Kludex/starlette
+                     (with anyio, MIT)
+  keyring            MIT          https://github.com/jaraco/keyring
+                     (with jaraco.classes, jaraco.context, jaraco.functools and
+                     more-itertools, MIT; on Linux also SecretStorage, BSD-3-Clause,
+                     jeepney, MIT, and cryptography, Apache-2.0 OR BSD-3-Clause, with
+                     cffi, MIT, and pycparser, BSD-3-Clause)
+
+Model files are downloaded from Hugging Face by the user and keep their own licenses; the
+Brasskey catalog stores only metadata (names, revisions, sizes, hashes), never weights.
+
+-------------------------------------------------------------------------------
+Brand assets
+-------------------------------------------------------------------------------
+
+The Brasskey logo outlines in brand/ are derived from Instrument Sans, licensed under the
+SIL Open Font License 1.1 (see brand/licenses/OFL-InstrumentSans.txt). The Brasskey name and
+logo are trademarks of Prcept AI LLP; see TRADEMARKS.md.

brasskey-0.1.0a1/PKG-INFO

@@ -0,0 +1,150 @@
+Metadata-Version: 2.5
+Name: brasskey
+Version: 0.1.0a1
+Summary: The key to every engine: deploy, tune, verify and serve open models on any hardware.
+Project-URL: Homepage, https://github.com/Prcept-AI/brasskey
+Project-URL: Repository, https://github.com/Prcept-AI/brasskey
+Project-URL: Issues, https://github.com/Prcept-AI/brasskey/issues
+Project-URL: Security, https://github.com/Prcept-AI/brasskey/blob/main/SECURITY.md
+Author: Prcept AI LLP
+License-Expression: Apache-2.0
+License-File: LICENSE
+License-File: NOTICE
+Keywords: benchmark,inference,llama.cpp,llm,mlx,openai-compatible,sglang,vllm
+Classifier: Development Status :: 2 - Pre-Alpha
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: System Administrators
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Typing :: Typed
+Requires-Python: >=3.11
+Requires-Dist: httpx2>=2.16
+Requires-Dist: huggingface-hub>=2.4
+Requires-Dist: keyring>=25.7
+Requires-Dist: nvidia-ml-py>=13.615
+Requires-Dist: platformdirs>=4.14
+Requires-Dist: psutil>=7.2
+Requires-Dist: pydantic>=2.14
+Requires-Dist: pyyaml>=6.0.3
+Requires-Dist: rich>=15.0
+Requires-Dist: starlette>=1.10
+Requires-Dist: typer>=0.27
+Requires-Dist: uvicorn>=0.54
+Description-Content-Type: text/markdown
+
+<br>
+
+<p align="center">
+  <picture>
+    <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/Prcept-AI/brasskey/v0.1.0a1/brand/logo/brasskey-wordmark-on-dark.svg">
+    <img alt="Brasskey" src="https://raw.githubusercontent.com/Prcept-AI/brasskey/v0.1.0a1/brand/logo/brasskey-wordmark-on-light.svg" width="240">
+  </picture>
+</p>
+
+<p align="center"><b>The key to every engine.</b></p>
+
+<br>
+
+Brasskey is an open-source tool to deploy, tune, verify and serve open models on any hardware,
+measured on your own traffic.
+
+```bash
+brasskey run        # try any model on your machine
+brasskey switch     # prove an open model can replace an expensive API (coming soon)
+brasskey deploy     # ship it tuned, verified and secure (coming soon)
+```
+
+## Status
+
+Pre-alpha. 0.1.0a1 is the first release on PyPI, on the way to v0.1 "It runs": `brasskey run`,
+`brasskey bench` and `brasskey doctor --security` on NVIDIA GPUs, Apple Silicon and CPUs, with
+llama.cpp, MLX, vLLM and SGLang. Done so far:
+
+- `brasskey doctor`: hardware, engines and running servers.
+- Planning with `brasskey run --dry-run`.
+- `brasskey run` with llama.cpp, MLX (Apple Silicon), vLLM or SGLang (NVIDIA on Linux, as a
+  process or from the official image pinned by digest). It downloads and verifies the model,
+  then serves it on 127.0.0.1 through a Brasskey proxy that enforces API keys. Ctrl-C stops it
+  cleanly. llama.cpp and MLX are tested on real hardware; vLLM and SGLang are not yet tested on
+  real GPUs.
+- `brasskey doctor --security`: checks this machine, or a URL, for exposed model servers.
+- `brasskey bench`: repeated GuideLLM runs, 95% intervals on every metric, and PASS / FAIL /
+  INCONCLUSIVE when comparing two setups.
+
+Next: testing vLLM and SGLang on NVIDIA GPUs, then 0.1.0. Design decisions are recorded in
+[docs/decisions](https://github.com/Prcept-AI/brasskey/tree/v0.1.0a1/docs/decisions/).
+
+## Install
+
+Python 3.11 or newer, on macOS or Linux (Windows through WSL2):
+
+```bash
+uv tool install brasskey       # or: pip install brasskey
+brasskey doctor                # what Brasskey sees on this machine
+brasskey run --dry-run         # the model it would pick for this machine, and why
+```
+
+`brasskey run` starts an engine you have installed, for example `brew install llama.cpp`; on
+Apple Silicon, Brasskey can set up MLX itself if you have uv. `brasskey doctor` shows what's
+installed and how to add the rest. Releases are signed and reproducible; see
+[SECURITY.md](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/SECURITY.md#verifying-a-release).
+
+## Try it from source
+
+You need [uv](https://docs.astral.sh/uv/) 0.13.6 or newer.
+
+```bash
+git clone https://github.com/Prcept-AI/brasskey.git
+cd brasskey
+uv sync
+uv run brasskey --help
+uv run brasskey doctor          # what Brasskey sees on this machine (CPU, memory, GPUs)
+uv run brasskey doctor --json
+uv run brasskey run --dry-run           # the model Brasskey would pick for this machine, and why
+uv run brasskey run qwen3-14b --dry-run --context 32768
+uv run brasskey run hf:Qwen/Qwen3-4B-GGUF:Q8_0 --dry-run   # any Hugging Face repo
+
+brew install llama.cpp                  # the engine (macOS or Linux)
+uv run brasskey run qwen3-0.6b          # download (639 MB), start, serve; Ctrl-C stops it
+uv run brasskey run qwen3-4b --save     # also writes brasskey.yaml + brasskey.lock
+uv run brasskey run --engine mlx        # Apple Silicon: MLX via uv, set up on first use
+uv run brasskey run --engine vllm       # NVIDIA on Linux: vLLM on PATH, or its pinned image
+
+uv run brasskey bench http://127.0.0.1:8000/v1 --save base.json      # measure a server
+uv run brasskey bench http://127.0.0.1:8001/v1 --baseline base.json  # PASS / FAIL / INCONCLUSIVE
+uv run brasskey bench http://127.0.0.1:8001/v1 --compare http://127.0.0.1:8000/v1
+
+uv run brasskey doctor --security       # exposed model servers, keys, Docker, firewall
+uv run brasskey doctor --security http://gpu-box:8000/v1   # check a server you run elsewhere
+uv run brasskey run qwen3-4b --host 0.0.0.0 --generate-api-key   # reachable from your network
+```
+
+Every command supports `--json` (one JSON document on stdout), `--debug` (logs and full
+tracebacks on stderr) and `--yes` (skip confirmations), and ends with a `Next:` suggestion.
+
+## Secure and private by default
+
+Servers bind to `127.0.0.1`. Any other address requires a strong API key (from
+`BRASSKEY_API_KEY` or your OS keychain, never a command-line argument). There is no telemetry,
+and prompt content is never logged. See [SECURITY.md](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/SECURITY.md).
+
+## Contributing
+
+See [CONTRIBUTING.md](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/CONTRIBUTING.md) and our [Code of Conduct](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/CODE_OF_CONDUCT.md).
+
+## License and pledge
+
+Apache-2.0 (see [LICENSE](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/LICENSE) and [NOTICE](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/NOTICE)). **The Brasskey core is Apache 2.0 and
+stays Apache 2.0.** Paid features for organizations live in separate repositories and never
+change the license of this one. The Brasskey name and logo are trademarks of Prcept AI LLP; see
+[TRADEMARKS.md](https://github.com/Prcept-AI/brasskey/blob/v0.1.0a1/TRADEMARKS.md).
+
+Made by Prcept AI LLP.

brasskey-0.1.0a1/README.md

@@ -0,0 +1,108 @@
+<br>
+
+<p align="center">
+  <picture>
+    <source media="(prefers-color-scheme: dark)" srcset="brand/logo/brasskey-wordmark-on-dark.svg">
+    <img alt="Brasskey" src="brand/logo/brasskey-wordmark-on-light.svg" width="240">
+  </picture>
+</p>
+
+<p align="center"><b>The key to every engine.</b></p>
+
+<br>
+
+Brasskey is an open-source tool to deploy, tune, verify and serve open models on any hardware,
+measured on your own traffic.
+
+```bash
+brasskey run        # try any model on your machine
+brasskey switch     # prove an open model can replace an expensive API (coming soon)
+brasskey deploy     # ship it tuned, verified and secure (coming soon)
+```
+
+## Status
+
+Pre-alpha. 0.1.0a1 is the first release on PyPI, on the way to v0.1 "It runs": `brasskey run`,
+`brasskey bench` and `brasskey doctor --security` on NVIDIA GPUs, Apple Silicon and CPUs, with
+llama.cpp, MLX, vLLM and SGLang. Done so far:
+
+- `brasskey doctor`: hardware, engines and running servers.
+- Planning with `brasskey run --dry-run`.
+- `brasskey run` with llama.cpp, MLX (Apple Silicon), vLLM or SGLang (NVIDIA on Linux, as a
+  process or from the official image pinned by digest). It downloads and verifies the model,
+  then serves it on 127.0.0.1 through a Brasskey proxy that enforces API keys. Ctrl-C stops it
+  cleanly. llama.cpp and MLX are tested on real hardware; vLLM and SGLang are not yet tested on
+  real GPUs.
+- `brasskey doctor --security`: checks this machine, or a URL, for exposed model servers.
+- `brasskey bench`: repeated GuideLLM runs, 95% intervals on every metric, and PASS / FAIL /
+  INCONCLUSIVE when comparing two setups.
+
+Next: testing vLLM and SGLang on NVIDIA GPUs, then 0.1.0. Design decisions are recorded in
+[docs/decisions](docs/decisions/).
+
+## Install
+
+Python 3.11 or newer, on macOS or Linux (Windows through WSL2):
+
+```bash
+uv tool install brasskey       # or: pip install brasskey
+brasskey doctor                # what Brasskey sees on this machine
+brasskey run --dry-run         # the model it would pick for this machine, and why
+```
+
+`brasskey run` starts an engine you have installed, for example `brew install llama.cpp`; on
+Apple Silicon, Brasskey can set up MLX itself if you have uv. `brasskey doctor` shows what's
+installed and how to add the rest. Releases are signed and reproducible; see
+[SECURITY.md](SECURITY.md#verifying-a-release).
+
+## Try it from source
+
+You need [uv](https://docs.astral.sh/uv/) 0.13.6 or newer.
+
+```bash
+git clone https://github.com/Prcept-AI/brasskey.git
+cd brasskey
+uv sync
+uv run brasskey --help
+uv run brasskey doctor          # what Brasskey sees on this machine (CPU, memory, GPUs)
+uv run brasskey doctor --json
+uv run brasskey run --dry-run           # the model Brasskey would pick for this machine, and why
+uv run brasskey run qwen3-14b --dry-run --context 32768
+uv run brasskey run hf:Qwen/Qwen3-4B-GGUF:Q8_0 --dry-run   # any Hugging Face repo
+
+brew install llama.cpp                  # the engine (macOS or Linux)
+uv run brasskey run qwen3-0.6b          # download (639 MB), start, serve; Ctrl-C stops it
+uv run brasskey run qwen3-4b --save     # also writes brasskey.yaml + brasskey.lock
+uv run brasskey run --engine mlx        # Apple Silicon: MLX via uv, set up on first use
+uv run brasskey run --engine vllm       # NVIDIA on Linux: vLLM on PATH, or its pinned image
+
+uv run brasskey bench http://127.0.0.1:8000/v1 --save base.json      # measure a server
+uv run brasskey bench http://127.0.0.1:8001/v1 --baseline base.json  # PASS / FAIL / INCONCLUSIVE
+uv run brasskey bench http://127.0.0.1:8001/v1 --compare http://127.0.0.1:8000/v1
+
+uv run brasskey doctor --security       # exposed model servers, keys, Docker, firewall
+uv run brasskey doctor --security http://gpu-box:8000/v1   # check a server you run elsewhere
+uv run brasskey run qwen3-4b --host 0.0.0.0 --generate-api-key   # reachable from your network
+```
+
+Every command supports `--json` (one JSON document on stdout), `--debug` (logs and full
+tracebacks on stderr) and `--yes` (skip confirmations), and ends with a `Next:` suggestion.
+
+## Secure and private by default
+
+Servers bind to `127.0.0.1`. Any other address requires a strong API key (from
+`BRASSKEY_API_KEY` or your OS keychain, never a command-line argument). There is no telemetry,
+and prompt content is never logged. See [SECURITY.md](SECURITY.md).
+
+## Contributing
+
+See [CONTRIBUTING.md](CONTRIBUTING.md) and our [Code of Conduct](CODE_OF_CONDUCT.md).
+
+## License and pledge
+
+Apache-2.0 (see [LICENSE](LICENSE) and [NOTICE](NOTICE)). **The Brasskey core is Apache 2.0 and
+stays Apache 2.0.** Paid features for organizations live in separate repositories and never
+change the license of this one. The Brasskey name and logo are trademarks of Prcept AI LLP; see
+[TRADEMARKS.md](TRADEMARKS.md).
+
+Made by Prcept AI LLP.