knaif 1.0.0__tar.gz

knaif-1.0.0/LICENSE

@@ -0,0 +1,201 @@
+                                 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 2026 knaif contributors
+
+   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.

knaif-1.0.0/NOTICE

@@ -0,0 +1,62 @@
+knaif
+Copyright 2026 Blackdeep Technologies
+
+This product includes software developed by Blackdeep Technologies.
+
+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 in the accompanying LICENSE file or at:
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+================================================================================
+MODELS
+================================================================================
+
+The fine-tuned models distributed alongside knaif are derivative works of the
+Qwen3 model family by Alibaba Cloud, used under the Apache License 2.0:
+
+    knaif-qwen3-4b-v1     derived from  Qwen/Qwen3-4B
+    knaif-qwen3-1.7b-v1   derived from  Qwen/Qwen3-1.7B
+
+    Qwen3 — Copyright Alibaba Cloud, licensed under Apache-2.0
+    https://huggingface.co/Qwen
+
+Models are NOT bundled in knaif artifacts; they are downloaded on first run from
+the URLs recorded in contracts/models/model-manifest.yaml, which also records each
+model's `base_model` and `base_model_license`.
+
+================================================================================
+BUNDLED THIRD-PARTY SOFTWARE
+================================================================================
+
+The native runtime links or bundles the following. Full license texts ship in
+the installers/licenses/ directory of every artifact.
+
+  llama.cpp — Copyright (c) 2023-2024 The ggml authors, MIT License
+      installers/licenses/llama.cpp-LICENSE.txt
+
+  NVIDIA CUDA runtime (CUDA-enabled builds only) — NVIDIA Software License
+  Agreement. CUDA support is an opt-in build; base artifacts do not include it.
+      installers/licenses/NVIDIA-CUDA-EULA.txt
+
+  Rust crate dependencies — permissive licenses only (Apache-2.0, MIT, BSD-2/3,
+  ISC, Unicode-3.0, NCSA, CDLA-Permissive-2.0). Generated by `just licenses`.
+      installers/licenses/THIRD-PARTY-RUST.txt
+
+  Python runtime dependencies — permissive licenses only (Apache-2.0, MIT,
+  BSD, MPL-2.0). Generated by `just licenses-python`.
+      installers/licenses/THIRD-PARTY-PYTHON.txt
+
+knaif does not bundle FFmpeg. Skills that shell out to FFmpeg or other external
+binaries require the user to install them separately; those tools remain under
+their own licenses. See skills/<name>/SPEC.md for per-skill requirements.
+
+================================================================================
+DATA
+================================================================================
+
+Evaluation, training, and safety corpora under skills/<name>/data/ were authored
+for this project and are covered by the project license. They reference synthetic
+filenames only and contain no third-party or personal content. See
+docs/PROVENANCE.md for the full non-code provenance record.

knaif-1.0.0/PKG-INFO

@@ -0,0 +1,315 @@
+Metadata-Version: 2.4
+Name: knaif
+Version: 1.0.0
+Summary: Local LLM command agent with deterministic execution layer
+Author-email: Blackdeep Technologies <knaif@blackdeep.tech>
+License-Expression: Apache-2.0
+Project-URL: Homepage, https://github.com/blackdeep-tech/knaif
+Project-URL: Repository, https://github.com/blackdeep-tech/knaif
+Project-URL: Documentation, https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/SDK.md
+Project-URL: Changelog, https://github.com/blackdeep-tech/knaif/blob/HEAD/CHANGELOG.md
+Project-URL: Issues, https://github.com/blackdeep-tech/knaif/issues
+Keywords: llm,cli,natural-language,agent,local-llm,ffmpeg
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: Microsoft :: Windows
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Application Frameworks
+Classifier: Topic :: Utilities
+Requires-Python: <3.15,>=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+License-File: NOTICE
+Requires-Dist: requests>=2.33.0
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: click>=8.0
+Provides-Extra: llama
+Requires-Dist: llama-cpp-python>=0.2.0; extra == "llama"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.4.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.1.0; extra == "dev"
+Requires-Dist: black>=23.7.0; extra == "dev"
+Requires-Dist: black[jupyter]>=26.5.1; extra == "dev"
+Requires-Dist: ruff>=0.0.280; extra == "dev"
+Requires-Dist: mypy>=1.5.0; extra == "dev"
+Requires-Dist: types-requests>=2.31.0; extra == "dev"
+Provides-Extra: notebook
+Requires-Dist: jupyterlab>=4.0.0; extra == "notebook"
+Requires-Dist: ipykernel>=6.25.0; extra == "notebook"
+Requires-Dist: notebook>=7.0.0; extra == "notebook"
+Requires-Dist: ipywidgets>=8.0.0; extra == "notebook"
+Provides-Extra: evalsuite
+Requires-Dist: rich>=13.0.0; extra == "evalsuite"
+Provides-Extra: clock
+Requires-Dist: tzdata>=2024.1; extra == "clock"
+Dynamic: license-file
+
+<p align="center">
+  <img src="https://raw.githubusercontent.com/blackdeep-tech/knaif/HEAD/media/knaif-logo-rect.svg" alt="knaif" width="260">
+</p>
+
+# knaif
+
+**Turn natural language into validated, executable actions — without letting a language
+model near your filesystem.**
+
+knaif runs a **local** model. No API key, no token meter, nothing leaves the machine.
+
+```python
+import knaif.cli as nk
+
+@nk.command(help="Add a task", keywords=["add", "create"])
+def add(title: nk.Arg(help="task title")):
+    ...
+
+nk.App([add]).run()      # your CLI now accepts plain English
+```
+
+## Why knaif
+
+Most ways of putting an LLM in front of a tool give the model the ability to *act* — it
+emits a shell command, or calls tools in a loop, and you hope the guardrails hold.
+
+knaif inverts that. The model is allowed to produce exactly one thing — a **plan**:
+
+```json
+{ "plan": [ { "tool": "convert", "args": { "input": "clip.mp4", "format": "webm" } } ] }
+```
+
+Everything that decides whether that plan runs is **ordinary deterministic code**: schema
+validation, arguments checked against a tool registry, sandbox path checks, a safety
+category per tool, and a confirmation gate. A hallucinated tool name is rejected before
+execution. An argument your tool doesn't accept is rejected. A path outside the sandbox is
+rejected — and re-checked after variable substitution, because substitution can introduce a
+new path.
+
+**What that buys you as a developer:**
+
+- **Safety is a property of your registry, not the model's mood.** You classify a tool
+  `destructive` once; the gate is code. Swap the model and the guarantee holds.
+- **One inference call, not an agent loop.** The model proposes the whole plan; everything
+  after is normal Python. No per-step round trip, no token meter, no nondeterminism in the
+  middle of a workflow.
+- **It runs on your user's hardware.** A small local model plus deterministic code means the
+  marginal cost of the millionth run equals the first — electricity on a laptop.
+- **A typed, testable surface.** Tools are declared in YAML and implemented as classes, so
+  the whole thing is unit-testable without calling a model at all — mock inference is built
+  in.
+
+Scope is narrow on purpose: knaif is not a general problem solver. It targets tools that are
+genuinely hard to drive by hand. **Free-form shell command generation is explicitly out of
+scope** — knaif will not emit a command its own validator can't account for.
+
+## Install
+
+```console
+pip install knaif
+```
+
+That's everything you need for mock inference and for Ollama. For in-process llama.cpp
+there is an extra:
+
+```console
+pip install "knaif[llama]"
+```
+
+> **Heads-up:** `llama-cpp-python` publishes only an sdist to PyPI, so this compiles from
+> source and needs CMake and a C++ toolchain. If you'd rather not build it, use the
+> project's [prebuilt wheels][llama-wheels] —
+> `pip install llama-cpp-python --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu`
+> — or just use Ollama, which needs no build step at all.
+
+## Try it without writing anything
+
+A complete example CLI ships inside the package, so you can see the whole pipeline work
+before writing a line:
+
+```console
+pip install "knaif[clock]"
+python -m knaif.examples.clock "list timezones in europe"
+python -m knaif.examples.clock "what time is it in Tokyo"
+```
+
+With no model installed it uses mock inference and still runs end to end — parse, validate,
+dispatch, result. Add Ollama or a GGUF (below) and the same command starts answering
+free-form phrasing. The source is ~340 lines across three files
+(`knaif/examples/clock/`): pure logic in `store.py`, then the same CLI built two ways —
+`@nk.command` decorators and `from_click()`.
+
+## Getting a model
+
+**No model is bundled** — not in this wheel, not anywhere. knaif is the engine; you choose
+what runs behind it. Three options, easiest first:
+
+**1. Ollama** — it manages the download for you, and needs no build step:
+
+```console
+ollama pull qwen3:4b
+```
+
+```python
+import knaif.cli as nk
+
+app = nk.App([convert], orchestrator=nk.local_ollama("qwen3:4b"))
+```
+
+`local_ollama()` returns `None` and warns if Ollama isn't reachable, so your app degrades to
+mock inference instead of crashing. Its defaults are also tuned for reasoning models like
+Qwen3 — use the helper rather than building an orchestrator by hand, or you'll hit a
+JSON-constraint deadlock. Details: [Reasoning models on Ollama][ollama-thinking].
+
+**2. A GGUF through llama.cpp** — one call, no llama.cpp knowledge required:
+
+```python
+app = nk.App([convert], orchestrator=nk.local_llama_cpp("model.gguf"))
+```
+
+`local_llama_cpp()` picks working defaults (8k context, full GPU offload, JSON handling for
+thinking-style models) so you don't have to learn llama.cpp's knobs. Override any of them —
+`n_ctx`, `n_gpu_layers`, `n_threads` — only if you want to. Like `local_ollama()`, it warns
+and returns `None` when the model or `llama-cpp-python` is missing, so your app falls back to
+mock rather than crashing.
+
+Any instruct-tuned GGUF that emits clean JSON works. Ready-made ones are at
+**[huggingface.co/blackdeep/knaif][hf]** — `knaif-qwen3-4b-v1` (~2.5 GB) is the
+general-purpose choice, `knaif-qwen3-1.7b-v1` is smaller and faster:
+
+```console
+curl -L -o model.gguf https://huggingface.co/blackdeep/knaif/resolve/main/knaif-qwen3-4b-v1-q4_k_m.gguf
+```
+
+A relative path resolves from your project root (the nearest directory containing `.git`,
+else the current directory); an absolute path always works. The warning names the path it
+actually tried, so a miss is easy to diagnose.
+
+**3. Mock inference** — the default when no orchestrator is supplied. It plans from registry
+keywords rather than a model, which makes the whole pipeline testable in CI with no model,
+no GPU, and no network.
+
+> The `knaif models pull` command and the `models.yaml` resolver belong to the **native CLI
+> and repo checkout**, not this package. From a pip install you supply the model explicitly,
+> as above.
+
+## Two ways to use it
+
+### 1. Add a natural-language front end to your own CLI
+
+The **self-contained** case: everything you need is in this package.
+
+```python
+import knaif.cli as nk
+
+@nk.command(help="Convert a file", keywords=["convert", "transcode"])
+def convert(src: nk.Arg(help="input file"), fmt: nk.Arg(help="target format")):
+    ...
+
+app = nk.App([convert])
+app.run()                                          # sys.argv[1] as an utterance
+app.invoke("turn report.docx into a pdf", dry_run=True)
+```
+
+Already using `click`? Wrap the existing group without touching your commands:
+
+```python
+from myapp.cli import cli
+app = nk.from_click(cli)
+```
+
+You get typed validation, safety categories, dry-run, and variable binding for free — the
+same pipeline the full engine uses.
+
+### 2. Build a skill
+
+When your domain outgrows a handful of commands, package it as a **skill** — a
+self-contained bundle of a YAML tool registry plus Python handlers. Skills give you intent
+expansion (one user request becoming a multi-step workflow), per-tool safety categories, and
+a reusable unit you can ship or hand to someone else.
+
+A minimal skill is a directory with `skill.yaml`, `tools.yaml`, and a handlers module:
+
+```python
+# my_skill/python/handlers.py
+from knaif.handler_api import HandlerContext
+from knaif.skill_base import Skill
+from knaif.tool import Step
+
+class GreetStep(Step):
+    name = "greet"
+    def handle(self, args: dict, ctx: HandlerContext) -> dict:
+        return {"message": f"hello {args['name']}"}
+
+class MySkill(Skill):
+    tools = [GreetStep]
+```
+
+Then load and run it:
+
+```python
+from knaif import CommandAgent
+
+agent = CommandAgent.from_skill("my_skill", sandbox="./sandbox")
+results = agent.execute_plan(
+    {"plan": [{"tool": "greet", "args": {"name": "world"}}]},
+    dry_run=True,
+)
+```
+
+Every tool is a `Step` (executes, returns a result) or an `Intent` (expands into a sub-plan
+of Steps). `HandlerContext` carries the sandbox root, `dry_run`, `confirmed`, and a
+confirmation callback. The full contract — registry schema, safety categories, intent
+expansion, preflight, result formatting — is in [TOOL_SCHEMA.md][tool-schema].
+
+Keeping several skills side by side? Point `create_agent()` at the directory holding them:
+
+```python
+from knaif import create_agent, list_skills
+
+print(list_skills(skills_root="./skills"))
+agent = create_agent("my_skill", sandbox="./sandbox", skills_root="./skills")
+```
+
+> **No skills ship in this wheel** — the published package is the engine plus the
+> `knaif.cli` SDK, so nothing pulls in dependencies for domains you don't use. If you want
+> worked examples rather than a blank page, the repository carries two production skills
+> (media processing and document handling) you can read or copy from.
+
+## Requirements
+
+- Python 3.10–3.14
+- A local model for real inference (llama.cpp via the `llama` extra, or Ollama). Mock
+  inference needs nothing.
+- Skills that shell out to external tools need those tools installed — external binaries are
+  never bundled.
+
+## Documentation
+
+| | |
+|---|---|
+| [Skill authoring][tool-schema] | registry schema, `Step` / `Intent`, safety categories |
+| [Developer SDK][sdk] | `@nk.command`, `nk.App`, `from_click` |
+| [Architecture][architecture] | the plan pipeline end to end |
+| [Inference backends][inference] | model resolution, llama.cpp and Ollama setup |
+| [Released models][models] | which models, why those, how they were tuned |
+| [Scope and safety][requirements] | what knaif deliberately will not do |
+
+Source, issues, and releases: **[github.com/blackdeep-tech/knaif][repo]**
+
+## License
+
+Apache-2.0 — see [LICENSE][license] and [NOTICE][notice].
+
+[repo]: https://github.com/blackdeep-tech/knaif
+[tool-schema]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/TOOL_SCHEMA.md
+[sdk]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/SDK.md
+[architecture]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/ARCHITECTURE.md
+[inference]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/INFERENCE.md
+[requirements]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/REQUIREMENTS.md
+[license]: https://github.com/blackdeep-tech/knaif/blob/HEAD/LICENSE
+[notice]: https://github.com/blackdeep-tech/knaif/blob/HEAD/NOTICE
+[models]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/MODELS.md
+[hf]: https://huggingface.co/blackdeep/knaif
+[ollama]: https://ollama.com
+[ollama-thinking]: https://github.com/blackdeep-tech/knaif/blob/HEAD/docs/INFERENCE.md#reasoning-models-on-ollama--leave-thinking-on
+[llama-wheels]: https://abetlen.github.io/llama-cpp-python/whl/cpu/