Skip to content
Back to search

tinker-skills

MarketplaceWorks with Claude CodeGitHub

Post-training with Tinker

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026. Its scripts were only partly scanned.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/thinking-machines-lab/tinker-cookbook with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/thinking-machines-lab/tinker-cookbook'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/thinking-machines-lab--tinker-cookbook/badge.svg)](/listings/thinking-machines-lab--tinker-cookbook)

HTML

<a href="/listings/thinking-machines-lab--tinker-cookbook"><img src="/api/listings/thinking-machines-lab--tinker-cookbook/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

1. Sign up for Tinker [here](https://auth.thinkingmachines.ai/sign-up). 2. Once you have access, create an API key from the [console](https://tinker-console.thinkingmachines.ai) and export it as environment variable `TINKER_API_KEY`. 3. Install `tinker-cookbook` (includes the `tinker` SDK as a dependency):

   # Latest stable release from PyPI
   uv pip install tinker-cookbook

   # Or install the nightly build
   uv pip install 'tinker-cookbook @ git+https://github.com/thinking-machines-lab/tinker-cookbook.git@nightly'

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 19 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "name": "tinker-skills",
  3. "owner": {
  4. "name": "Thinking Machines Lab",
  5. "email": "tinker@thinkingmachines.ai"
  6. },
  7. "metadata": {
  8. "description": "Claude Code skills for fine-tuning language models with the Tinker API.",
  9. "version": "0.3.0"
  10. },
  11. "plugins": [
  12. {
  13. "name": "tinker",
  14. "description": "Skills for fine-tuning language models with the Tinker API — research, debugging, and more.",
  15. "source": "./",
  16. "strict": true
  17. }
  18. ]
  19. }
README.md · README · 187 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <h1 align="center">Tinker Cookbook</h1>
  2. <div align="center">
  3. <img src="assets/tinker-cover.png" width="60%" />
  4. </div>
  5. <div align="center">
  6. [![pytest](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pytest.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pytest.yaml)
  7. [![pyright](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pyright.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/pyright.yaml)
  8. [![smoke-test-recipes](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/smoke-test-recipes.yaml/badge.svg)](https://github.com/thinking-machines-lab/tinker-cookbook/actions/workflows/smoke-test-recipes.yaml)
  9. [![PyPI](https://img.shields.io/pypi/v/tinker-cookbook)](https://pypi.org/project/tinker-cookbook/)
  10. </div>
  11. We provide two libraries for the broader community to customize their language models: `tinker` and `tinker-cookbook`.
  12. - `tinker` is a training SDK for researchers and developers to fine-tune language models. You send API requests to us and we handle the complexities of distributed training.
  13. - `tinker-cookbook` includes realistic examples of fine-tuning language models. It builds on the Tinker API and provides common abstractions to fine-tune language models.
  14. ## Installation
  15. 1. Sign up for Tinker [here](https://auth.thinkingmachines.ai/sign-up).
  16. 2. Once you have access, create an API key from the [console](https://tinker-console.thinkingmachines.ai) and export it as environment variable `TINKER_API_KEY`.
  17. 3. Install `tinker-cookbook` (includes the `tinker` SDK as a dependency):
  18. ```bash
  19. # Latest stable release from PyPI
  20. uv pip install tinker-cookbook
  21. # Or install the nightly build
  22. uv pip install 'tinker-cookbook @ git+https://github.com/thinking-machines-lab/tinker-cookbook.git@nightly'
  23. ```
  24. ## Using Inkling models
  25. Inkling is Thinking Machines Lab's model family tailored for Tinker:
  26. general-purpose models that can code, reason, and call tools. Inkling is also
  27. natively multimodal. Every Inkling model shares the same rendering path and
  28. reasoning-effort interface. Tinker Cookbook supports sampling and post-training
  29. every Inkling model through the standalone `tml-renderers` library, included in
  30. the default installation together with the required `torch>=2.10`.
  31. Pass `model_name="thinkingmachines/Inkling-2-Small"`,
  32. `model_name="thinkingmachines/Inkling"`, or
  33. `model_name="thinkingmachines/Inkling-Small"` anywhere the cookbook accepts a
  34. model name. The appropriate renderer and tokenizer are selected automatically for
  35. each of these models, including the `:peft:` long-context variants. The tokenizer
  36. depends on the model, so always get it from the model name.
  37. All Inkling models support [controllable reasoning effort](https://tinker-docs.thinkingmachines.ai/cookbook/inkling/thinking-effort/)
  38. and multimodal [image](https://tinker-docs.thinkingmachines.ai/cookbook/inkling/images/)
  39. inputs. Inkling and Inkling-Small also accept
  40. [audio](https://tinker-docs.thinkingmachines.ai/cookbook/inkling/audio/) inputs.
  41. See the [Inkling documentation](https://tinker-docs.thinkingmachines.ai/cookbook/inkling/)
  42. for setup and usage details, and
  43. [Models & Pricing](https://tinker-docs.thinkingmachines.ai/tinker/models/) for
  44. per-model context lengths and pricing.
  45. ## Tinker
  46. Here we introduce a few Tinker primitives — the basic components to fine-tune LLMs (see the [quickstart guide](https://tinker-docs.thinkingmachines.ai/tinker/quickstart/) for more details):
  47. ```python
  48. import tinker
  49. service_client = tinker.ServiceClient()
  50. training_client = service_client.create_lora_training_client(
  51. base_model="meta-llama/Llama-3.2-1B", rank=32,
  52. ).result()
  53. training_client.forward_backward(...)
  54. training_client.optim_step(...)
  55. training_client.save_state(...)
  56. resumed_client = service_client.create_training_client_from_state(...).result()
  57. sampling_client = training_client.save_weights_and_get_sampling_client().result()
  58. sampling_client.sample(...)
  59. ```
  60. See [tinker_cookbook/recipes/sl_loop.py](tinker_cookbook/recipes/sl_loop.py) and [tinker_cookbook/recipes/rl_loop.py](tinker_cookbook/recipes/rl_loop.py) for minimal examples of using these primitives to fine-tune LLMs.
  61. ### Tutorials
  62. New to Tinker? The [`tutorials/`](tutorials/) directory contains 20+ progressive [marimo](https://marimo.io/) notebooks that walk through core concepts — rendering, loss functions, completers, weight management — and advanced topics such as custom RL environments, DPO, RLHF, and weight export. Run any tutorial with `marimo edit tutorials/101_hello_tinker.py`. See the [tutorials README](tutorials/README.md) for the full list, or browse rendered versions on the [Tinker docs site](https://tinker-docs.thinkingmachines.ai/tutorials).
  63. To export and download trained weights:
  64. ```python
  65. from tinker_cookbook import weights
  66. export = training_client.save_weights_external(name="export").result()
  67. weights.download(tinker_path=export.path, output_dir="model-checkpoint")
  68. ```
  69. ### Tinker Cookbook
  70. Besides these primitives, we also offer **Tinker Cookbook** (a.k.a. this repo), a library of a wide range of abstractions to help you customize training environments.
  71. [`tinker_cookbook/recipes/sl_basic.py`](tinker_cookbook/recipes/sl_basic.py) and [`tinker_cookbook/recipes/rl_basic.py`](tinker_cookbook/recipes/rl_basic.py) contain minimal examples to configure supervised learning and reinforcement learning.
  72. We also include more complete examples in the [`tinker_cookbook/recipes/`](tinker_cookbook/recipes/) folder:
  73. - **[Chat SFT](tinker_cookbook/recipes/chat_sl/)**: supervised fine-tuning on conversational datasets (e.g., Tulu3).
  74. - **[Math RL](tinker_cookbook/recipes/math_rl/)**: reinforcement learning for mathematical reasoning with verifiable rewards.
  75. - **[Code RL](tinker_cookbook/recipes/code_rl/)**: RL on competitive programming with sandboxed code execution (DeepCoder replication).
  76. - **[Preference learning](tinker_cookbook/recipes/preference/)**: DPO and a three-stage RLHF pipeline (SFT, reward model, RL).
  77. - **[Distillation](tinker_cookbook/recipes/distillation/)**: on-policy and off-policy knowledge distillation with single- and multi-teacher configurations.
  78. - **[Tool use](tinker_cookbook/recipes/search_tool/)**: RL for retrieval-augmented generation (Search-R1 replication).
  79. - **[Multi-agent](tinker_cookbook/recipes/multiplayer_rl/)**: multi-agent RL with self-play and cross-play.
  80. - **[Audio](tinker_cookbook/recipes/audio/)**: SFT and RL for Inkling audio inputs, including speech recognition and speaking-style classification.
  81. - **[VLM image classification](tinker_cookbook/recipes/vlm_classifier/)**: supervised training and evaluation for vision-language models.
  82. The [recipes README](tinker_cookbook/recipes/README.md) covers all available recipes, including Harbor RL, rubric-based grading, VLM classification, and SDFT. Each recipe includes a `README.md` with implementation details, launch commands, and expected results.
  83. ### Evaluation (experimental)
  84. Tinker Cookbook includes a [benchmark framework](tinker_cookbook/eval/) for evaluating trained models:
  85. ```python
  86. from tinker_cookbook import cli_utils
  87. from tinker_cookbook.eval.benchmarks import run_benchmarks, BenchmarkConfig
  88. results = await run_benchmarks(
  89. ["gsm8k", "mmlu_pro", "ifeval"],
  90. sampling_client, renderer,
  91. BenchmarkConfig(save_dir=cli_utils.runs_path("evals", "step500")),
  92. )
  93. ```
  94. The framework currently supports 12 benchmarks (GSM8K, MATH-500, MMLU-Pro, MMLU-Redux, GPQA, IFEval, MBPP, C-Eval, SuperGPQA, IFBench, AIME 2025, AIME 2026) with verified scores against published results, plus experimental benchmarks such as LiveCodeBench, Terminal Bench, and SWE-bench. Benchmarks can also serve as inline training evaluators via `BenchmarkEvaluator`.
  95. **Note:** Benchmark scores are sensitive to evaluation configuration — system prompts, `max_tokens`, temperature, and timeout settings can shift results significantly. We document our exact settings alongside all reported scores. This framework is under active development; feedback and contributions are welcome. See the [eval README](tinker_cookbook/eval/README.md) for verified scores, configuration details, and instructions for adding new benchmarks.
  96. ### Documentation
  97. For the full Tinker documentation, visit [tinker-docs.thinkingmachines.ai](https://tinker-docs.thinkingmachines.ai).
  98. ### Utilities
  99. Tinker Cookbook also provides reusable building blocks:
  100. - [`renderers`](tinker_cookbook/renderers/) — bidirectional conversion between token sequences and structured chat messages
  101. - [`hyperparam_utils`](tinker_cookbook/hyperparam_utils.py) — calibrated per-model LoRA learning rates and LoRA parameter counts
  102. - [`eval`](tinker_cookbook/eval/) — benchmark framework and inline training evaluators (see [Evaluation](#evaluation-experimental) above)
  103. ## Claude Code Skills
  104. Tinker Cookbook ships with [Claude Code skills](https://docs.anthropic.com/en/docs/claude-code/skills) that teach Claude how to use the Tinker API. Install them so Claude can help you write training code in any project:
  105. ```
  106. /plugin marketplace add thinking-machines-lab/tinker-cookbook
  107. ```
  108. Then install the **tinker** plugin from the Discover tab (`/plugin` → Discover). Once installed, four skills are available:
  109. | Command | What it does |
  110. |---|---|
  111. | `/tinker:research` | Plan and run post-training experiments — SFT, RL, DPO, distillation, evaluation, hyperparameters, model selection, and more |
  112. | `/tinker:debug` | Diagnose slow training, hangs, output mismatches, renderer issues, and errors |
  113. | `/tinker:rollout-ui` | Build rollout viewers with messages, tool calls, provenance, aligned comparisons, and accessible diffs; link to W&B for metrics |
  114. | `/tinker:inkling` | Work with Inkling models, including Inkling 2 Small — thinking effort, rendering, post-training, evaluation, and multimodal input |
  115. Skills also trigger automatically based on context — ask Claude to "set up SFT training" and it will load the right skill without a slash command. Skills update automatically when the repo is updated.
  116. ## Development Setup
  117. ```bash
  118. uv sync --extra dev
  119. pre-commit install
  120. ```
  121. This installs dev dependencies and registers pre-commit hooks that run `ruff` formatting and linting on every commit. CI enforces these checks on all pull requests.
  122. ## Contributing
  123. This project is built in the spirit of open science and collaborative development. We believe that the best tools emerge through community involvement and shared learning.
  124. We welcome PR contributions after our private beta is over. If you have any feedback, please email us at tinker@thinkingmachines.ai.
  125. ## Citation
  126. If you use Tinker for your research, please cite it as:
  127. ```
  128. Thinking Machines Lab, 2026. Tinker. https://thinkingmachines.ai/tinker/.
  129. ```
  130. Or use this BibTeX citation:
  131. ```
  132. @misc{tml2026tinker,
  133. author = {Thinking Machines Lab},
  134. title = {Tinker},
  135. year = {2026},
  136. url = {https://thinkingmachines.ai/tinker/},
  137. }
  138. ```

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.