trainlint 0.1.0__tar.gz

trainlint-0.1.0/.claude/settings.local.json

@@ -0,0 +1,7 @@
+{
+    "attribution": {
+      "commit": "",
+      "pr": "",
+      "sessionUrl": false
+    }
+  }

trainlint-0.1.0/.gitattributes

@@ -0,0 +1 @@
+* text=auto eol=lf

trainlint-0.1.0/.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,23 @@
+name: Bug report
+description: Something other than a wrong finding is broken
+labels: ["bug"]
+body:
+  - type: textarea
+    id: what
+    attributes:
+      label: What happened
+      description: The command you ran, what you expected, and what you got.
+    validations:
+      required: true
+  - type: input
+    id: version
+    attributes:
+      label: Output of `trainlint --version`
+    validations:
+      required: true
+  - type: input
+    id: platform
+    attributes:
+      label: Operating system and Python version
+    validations:
+      required: true

trainlint-0.1.0/.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1 @@
+blank_issues_enabled: false

trainlint-0.1.0/.github/ISSUE_TEMPLATE/false_positive.yml

@@ -0,0 +1,30 @@
+name: False positive
+description: trainlint flagged code that is correct
+labels: ["false positive"]
+body:
+  - type: input
+    id: rule
+    attributes:
+      label: Rule
+      placeholder: TL001
+    validations:
+      required: true
+  - type: textarea
+    id: snippet
+    attributes:
+      label: Smallest snippet that is flagged
+      render: python
+    validations:
+      required: true
+  - type: textarea
+    id: why
+    attributes:
+      label: Why this code is correct
+    validations:
+      required: true
+  - type: input
+    id: version
+    attributes:
+      label: Output of `trainlint --version`
+    validations:
+      required: true

trainlint-0.1.0/.github/ISSUE_TEMPLATE/rule_idea.yml

@@ -0,0 +1,27 @@
+name: Rule idea
+description: A bug that trainlint should catch
+labels: ["new rule"]
+body:
+  - type: textarea
+    id: bug
+    attributes:
+      label: The bug
+      description: What goes wrong, and how you would notice it (or not).
+    validations:
+      required: true
+  - type: textarea
+    id: example
+    attributes:
+      label: Minimal example
+      render: python
+    validations:
+      required: true
+  - type: textarea
+    id: signal
+    attributes:
+      label: What in the code gives it away
+      description: |
+        Rules work on one file's syntax tree, with no type information. Name the calls, names or
+        ordering a check could look at, and a sensible script that looks similar but is fine.
+    validations:
+      required: true

trainlint-0.1.0/.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,8 @@
+## What changed
+
+## How I checked it
+
+- [ ] `uv run pytest` passes
+- [ ] `uv run ruff check . && uv run ruff format --check .` passes
+- [ ] New or changed rule: buggy and fixed files added under `bench/`, example added to `docs/rules.md`
+- [ ] New or changed rule: I read what it reports on at least one real codebase

trainlint-0.1.0/.github/workflows/ci.yml

@@ -0,0 +1,43 @@
+name: ci
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+
+jobs:
+  test:
+    name: test (${{ matrix.os }}, python ${{ matrix.python }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, windows-latest, macos-latest]
+        python: ["3.10", "3.13"]
+    steps:
+      - uses: actions/checkout@v4
+      - uses: astral-sh/setup-uv@v5
+      - run: uv sync --locked --python ${{ matrix.python }}
+      - run: uv run pytest -q
+      - run: uv run python -m bench
+
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: astral-sh/setup-uv@v5
+      - run: uv sync --locked
+      - run: uv run ruff check .
+      - run: uv run ruff format --check .
+
+  package:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: astral-sh/setup-uv@v5
+      - run: uv build
+      - name: Install the wheel into a clean environment and run it
+        run: |
+          uvx --from dist/*.whl trainlint skills install --dest "$RUNNER_TEMP/skills"
+          test -f "$RUNNER_TEMP/skills/ml-training-loop/SKILL.md"
+          uvx --from dist/*.whl trainlint check --exit-zero bench/buggy

trainlint-0.1.0/.gitignore

@@ -0,0 +1,8 @@
+__pycache__/
+*.py[cod]
+.venv/
+.pytest_cache/
+.ruff_cache/
+dist/
+build/
+*.egg-info/

trainlint-0.1.0/CHANGELOG.md

@@ -0,0 +1,14 @@
+# Changelog
+
+## 0.1.0 - 2026-11-03
+
+First release.
+
+- `trainlint check` with 12 rules (TL001 to TL012) for data leakage, train and eval mode,
+  seeding, mixed precision, checkpoints, loss inputs, gradient clearing, tensor accumulation,
+  metric misuse and evaluation loaders. Text and JSON output, `--select`, `--ignore`,
+  `--exit-zero`, `--color`, and `# trainlint: ignore[RULE]` comments.
+- Four Agent Skills (`ml-data-leakage`, `ml-training-loop`, `ml-metrics-eval`,
+  `ml-checkpoint-resume`) and `trainlint skills install`.
+- Regression corpus of 18 buggy scripts with fixed twins (`python -m bench`).
+- Per-rule reference in `docs/rules.md`.

trainlint-0.1.0/CONTRIBUTING.md

@@ -0,0 +1,46 @@
+# Contributing
+
+Reports of false positives and missed bugs are the most useful thing you can send. A rule that
+cries wolf gets switched off, so precision matters more here than coverage.
+
+## Set up
+
+You need [uv](https://docs.astral.sh/uv/) and Python 3.10 or newer.
+
+```console
+git clone https://github.com/Satvik-jain/trainlint
+cd trainlint
+uv sync
+uv run pytest
+uv run ruff check . && uv run ruff format --check .
+uv run python -m bench
+```
+
+## Reporting a problem
+
+- **False positive:** use the "False positive" issue form. Paste the smallest snippet that is
+  flagged and say why it is fine.
+- **Missed bug:** use the "Rule idea" form. Show the bug and say what in the code gives it away.
+  A rule has to work from one file's syntax tree, with no types and no imports resolved.
+
+## Adding or changing a rule
+
+1. Write the check as a function `(tree: ast.Module) -> Iterator[Hit]` in the module for its
+   topic under `src/trainlint/rules/`, and register it in that module's `RULES`. Helpers for
+   matching calls and names live in `src/trainlint/astutil.py`.
+2. Add unit tests with both flagged and clean snippets. Clean snippets matter more: write down
+   the ways a sensible script could look like the bug without being it.
+3. Add a buggy script and its fixed twin to `bench/buggy/` and `bench/fixed/`. The buggy file
+   starts with `# expect: TLxxx`.
+4. Add the rule to `docs/rules.md` with one flagged and one clean example. A test runs both.
+5. If a bundled skill should mention it, add it to the skill under `src/trainlint/skills/`.
+   A test checks that skills only cite rules that exist and that every rule is cited.
+
+A new rule should be quiet on ordinary code. If you can, run it over a few real repositories and
+read what it reports before opening the pull request.
+
+## Commits and pull requests
+
+Keep commits small and use Conventional Commit prefixes (`feat`, `fix`, `docs`, `test`, `chore`,
+`refactor`, `ci`), imperative mood, no trailing period. Say in the pull request what changed and
+how you checked it.

trainlint-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Satvik Jain
+
+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.

trainlint-0.1.0/PKG-INFO

@@ -0,0 +1,204 @@
+Metadata-Version: 2.5
+Name: trainlint
+Version: 0.1.0
+Summary: Offline checker and agent skills for ML bugs that do not crash: leakage, train/eval mode, loss and metric mistakes.
+Project-URL: Homepage, https://github.com/Satvik-jain/trainlint
+Project-URL: Issues, https://github.com/Satvik-jain/trainlint/issues
+Author: Satvik Jain
+License-Expression: MIT
+License-File: LICENSE
+Keywords: agent-skills,data-leakage,linter,machine-learning,pytorch,scikit-learn,static-analysis
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Quality Assurance
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+
+# trainlint
+
+Agent skills and an offline checker for ML bugs that do not crash.
+
+[![ci](https://github.com/Satvik-jain/trainlint/actions/workflows/ci.yml/badge.svg)](https://github.com/Satvik-jain/trainlint/actions/workflows/ci.yml)
+[![license: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+
+![Cross-validated accuracy on random labels: 85% with a leak, trainlint pointing at it, 48% after the fix](assets/demo.gif)
+
+The 5,000 features and the labels are random, so no model can beat 50%. Picking features before
+cross-validating reports 85%. The GIF is rendered from the real output of the commands shown;
+`assets/make_assets.py` rebuilds it.
+
+## Install
+
+```console
+uvx trainlint check .
+```
+
+or `pipx install trainlint` / `pip install trainlint`. It needs Python 3.10 or newer, has no
+dependencies, and never touches the network or a GPU.
+
+## Quickstart
+
+```console
+uvx trainlint check train.py     # find the bugs in a script, a folder, or a repo
+uvx trainlint skills install     # teach your coding agent to avoid them
+```
+
+`skills install` copies four skills into `./.claude/skills`. Use `--agent agents` for
+`./.agents/skills`, `--global` for your user folder, or `--dest DIR` for any other place.
+
+More ways to run the checker:
+
+```console
+trainlint check src/ --ignore TL005   # skip a rule
+trainlint check . --format json       # for scripts and CI
+trainlint rules                       # list every rule
+```
+
+Exit code 0 means no findings, 1 means findings, 2 means a usage error. To silence a finding on
+one line:
+
+```python
+scaler.fit(X_all)  # trainlint: ignore[TL002] fitting on the pooled set is intended here
+```
+
+In GitHub Actions:
+
+```yaml
+- run: uvx trainlint check .
+```
+
+## Why it exists
+
+ML code fails quietly. A scaler fitted before the split, a model left in eval mode, or a softmax
+in front of `CrossEntropyLoss` still trains and still prints a number. Coding agents and people
+both write these mistakes, and general linters do not know what the code is for. trainlint is a
+small offline checker for the common cases, plus skills that tell an agent how to avoid them in
+the first place.
+
+## What it does
+
+- Twelve rules, TL001 to TL012, listed below and explained with examples in
+  [docs/rules.md](docs/rules.md).
+- Each finding shows the line, a pointer into the code, and a one-line fix.
+- Pure standard library. It parses the code and never runs it.
+- Text or JSON output, rule selection, and per-line ignore comments.
+- Four [Agent Skills](https://agentskills.io/specification) that cover the same ground and run
+  the checker when it is installed.
+
+| Rule | Flags |
+| --- | --- |
+| TL001 | Scaler, encoder, selector or resampler fitted on all rows before a split or cross-validation |
+| TL002 | Transform or model fitted on a validation or test split |
+| TL003 | `model.eval()` inside a training loop with no `model.train()` to switch back |
+| TL004 | Evaluation under `no_grad` in a script that never calls `model.eval()` |
+| TL005 | Training script that never sets a random seed |
+| TL006 | float16 autocast without a `GradScaler`, or a scaled loss stepped without `scaler.step` |
+| TL007 | Script that resumes from checkpoints but saves no optimizer state |
+| TL008 | `softmax` or `sigmoid` at the end of a model trained with a logits loss |
+| TL009 | Gradients never cleared, or cleared between `backward()` and `step()` |
+| TL010 | Loss added to a running total or list as a tensor, keeping every graph alive |
+| TL011 | `y_true` and `y_pred` swapped, or class labels passed to `roc_auc_score` |
+| TL012 | Validation or test `DataLoader` with `drop_last=True` |
+
+| Skill | Covers |
+| --- | --- |
+| `ml-data-leakage` | Splits, pipelines, grouped and time-ordered data, resampling, the test set |
+| `ml-training-loop` | Modes, gradient order, loss inputs, shapes, seeds, mixed precision |
+| `ml-metrics-eval` | Argument order, scores versus labels, averaging, imbalance, variance |
+| `ml-checkpoint-resume` | What a checkpoint must hold, best versus last, verifying a resume |
+
+## How far to trust it
+
+The checks parse each file with Python's `ast` module and match calls by name. There is no type
+inference and no cross-file analysis, so:
+
+- Splits are recognized by variable names such as `X_train`, `val_loader` and `y_test`.
+- A bug whose cause sits in another file is not seen.
+- A heuristic can be wrong. When it is, add an ignore comment and open an issue.
+
+`python -m bench` runs the checker over a corpus of 18 buggy scripts, one or two per rule, each
+with a fixed twin:
+
+```
+rule    buggy files  caught   false alarms on fixed twins
+TL001             3       3                             0
+TL002             1       1                             0
+TL003             1       1                             0
+TL004             1       1                             0
+TL005             1       1                             0
+TL006             2       2                             0
+TL007             1       1                             0
+TL008             2       2                             0
+TL009             2       2                             0
+TL010             1       1                             0
+TL011             2       2                             0
+TL012             1       1                             0
+total            18      18                             0
+```
+
+I wrote that corpus together with the rules, so it guards against regressions and says nothing
+about accuracy on code I have not seen.
+
+For a check on someone else's code, I ran it over all 93 Python files of
+[pytorch/examples](https://github.com/pytorch/examples) at commit `acc295d` and read every
+finding:
+
+```console
+git clone https://github.com/pytorch/examples && cd examples
+git checkout acc295dc7b90714f1bf47f06004fc19a7fe235c4
+trainlint check .
+```
+
+It reports 21 findings. By my reading, 19 are real: 15 scripts never set a seed (one declares a
+`--seed` option and never uses it), 3 tensor-parallel scripts never clear gradients, and one script
+resumes from a snapshot without saving optimizer state. Two more are correct as written but
+harmless, because those models have no dropout or batch norm to leave in training mode. The first
+run also flagged three distributed-autograd scripts for missing `zero_grad()`, which was a false
+positive that is now fixed. This is one repository of tutorial code and I did the reviewing, so
+treat it as a spot check. It tells you nothing about what the rules miss.
+
+## Compared with other tools
+
+| Tool | What it covers | Where trainlint differs |
+| --- | --- | --- |
+| ruff, pylint, flake8 | General Python style and bugs | None of their rules know about splits, eval mode or metrics. Use them alongside trainlint. |
+| [TorchFix](https://github.com/meta-pytorch/torchfix) | Removed and deprecated PyTorch functions, `requires_grad` typos, unsafe `torch.load` | Its README lists no rules for training logic. trainlint does not check API deprecations. |
+| [mllint](https://github.com/bvobart/mllint) | Repository-level quality of ML projects: version control, dependencies, code quality | It looks at project hygiene. trainlint reads training and evaluation code. |
+| [Deepchecks](https://github.com/deepchecks/deepchecks) | Checks that run on real datasets and models, including drift | Deepchecks needs your data and model loaded. trainlint reads source only and runs nothing. |
+| General skill collections | Engineering process for coding agents | They are not specific to ML. trainlint's skills target training and evaluation code and come with a checker. |
+
+## Roadmap
+
+- Jupyter notebook support
+- SARIF output for GitHub code scanning
+- Settings in `pyproject.toml`
+- More rules: training loaders without shuffling, `retain_graph` in a loop, normalization on a
+  whole DataFrame before the split
+- A benchmark that measures how often an agent writes these bugs with and without the skills
+  (needs an API key, so it will be optional)
+
+Issues labeled [good first issue](https://github.com/Satvik-jain/trainlint/labels/good%20first%20issue)
+are small and have a clear definition of done.
+
+## Contributing
+
+False positives and missed bugs are the most useful reports. See
+[CONTRIBUTING.md](CONTRIBUTING.md) for setup and for how to add a rule.
+
+```console
+uv sync
+uv run pytest
+uv run ruff check . && uv run ruff format --check .
+uv run python -m bench
+```
+
+## License
+
+MIT
+
+If trainlint saved you an afternoon, a star on the repository helps other people find it.

trainlint-0.1.0/README.md

@@ -0,0 +1,183 @@
+# trainlint
+
+Agent skills and an offline checker for ML bugs that do not crash.
+
+[![ci](https://github.com/Satvik-jain/trainlint/actions/workflows/ci.yml/badge.svg)](https://github.com/Satvik-jain/trainlint/actions/workflows/ci.yml)
+[![license: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
+
+![Cross-validated accuracy on random labels: 85% with a leak, trainlint pointing at it, 48% after the fix](assets/demo.gif)
+
+The 5,000 features and the labels are random, so no model can beat 50%. Picking features before
+cross-validating reports 85%. The GIF is rendered from the real output of the commands shown;
+`assets/make_assets.py` rebuilds it.
+
+## Install
+
+```console
+uvx trainlint check .
+```
+
+or `pipx install trainlint` / `pip install trainlint`. It needs Python 3.10 or newer, has no
+dependencies, and never touches the network or a GPU.
+
+## Quickstart
+
+```console
+uvx trainlint check train.py     # find the bugs in a script, a folder, or a repo
+uvx trainlint skills install     # teach your coding agent to avoid them
+```
+
+`skills install` copies four skills into `./.claude/skills`. Use `--agent agents` for
+`./.agents/skills`, `--global` for your user folder, or `--dest DIR` for any other place.
+
+More ways to run the checker:
+
+```console
+trainlint check src/ --ignore TL005   # skip a rule
+trainlint check . --format json       # for scripts and CI
+trainlint rules                       # list every rule
+```
+
+Exit code 0 means no findings, 1 means findings, 2 means a usage error. To silence a finding on
+one line:
+
+```python
+scaler.fit(X_all)  # trainlint: ignore[TL002] fitting on the pooled set is intended here
+```
+
+In GitHub Actions:
+
+```yaml
+- run: uvx trainlint check .
+```
+
+## Why it exists
+
+ML code fails quietly. A scaler fitted before the split, a model left in eval mode, or a softmax
+in front of `CrossEntropyLoss` still trains and still prints a number. Coding agents and people
+both write these mistakes, and general linters do not know what the code is for. trainlint is a
+small offline checker for the common cases, plus skills that tell an agent how to avoid them in
+the first place.
+
+## What it does
+
+- Twelve rules, TL001 to TL012, listed below and explained with examples in
+  [docs/rules.md](docs/rules.md).
+- Each finding shows the line, a pointer into the code, and a one-line fix.
+- Pure standard library. It parses the code and never runs it.
+- Text or JSON output, rule selection, and per-line ignore comments.
+- Four [Agent Skills](https://agentskills.io/specification) that cover the same ground and run
+  the checker when it is installed.
+
+| Rule | Flags |
+| --- | --- |
+| TL001 | Scaler, encoder, selector or resampler fitted on all rows before a split or cross-validation |
+| TL002 | Transform or model fitted on a validation or test split |
+| TL003 | `model.eval()` inside a training loop with no `model.train()` to switch back |
+| TL004 | Evaluation under `no_grad` in a script that never calls `model.eval()` |
+| TL005 | Training script that never sets a random seed |
+| TL006 | float16 autocast without a `GradScaler`, or a scaled loss stepped without `scaler.step` |
+| TL007 | Script that resumes from checkpoints but saves no optimizer state |
+| TL008 | `softmax` or `sigmoid` at the end of a model trained with a logits loss |
+| TL009 | Gradients never cleared, or cleared between `backward()` and `step()` |
+| TL010 | Loss added to a running total or list as a tensor, keeping every graph alive |
+| TL011 | `y_true` and `y_pred` swapped, or class labels passed to `roc_auc_score` |
+| TL012 | Validation or test `DataLoader` with `drop_last=True` |
+
+| Skill | Covers |
+| --- | --- |
+| `ml-data-leakage` | Splits, pipelines, grouped and time-ordered data, resampling, the test set |
+| `ml-training-loop` | Modes, gradient order, loss inputs, shapes, seeds, mixed precision |
+| `ml-metrics-eval` | Argument order, scores versus labels, averaging, imbalance, variance |
+| `ml-checkpoint-resume` | What a checkpoint must hold, best versus last, verifying a resume |
+
+## How far to trust it
+
+The checks parse each file with Python's `ast` module and match calls by name. There is no type
+inference and no cross-file analysis, so:
+
+- Splits are recognized by variable names such as `X_train`, `val_loader` and `y_test`.
+- A bug whose cause sits in another file is not seen.
+- A heuristic can be wrong. When it is, add an ignore comment and open an issue.
+
+`python -m bench` runs the checker over a corpus of 18 buggy scripts, one or two per rule, each
+with a fixed twin:
+
+```
+rule    buggy files  caught   false alarms on fixed twins
+TL001             3       3                             0
+TL002             1       1                             0
+TL003             1       1                             0
+TL004             1       1                             0
+TL005             1       1                             0
+TL006             2       2                             0
+TL007             1       1                             0
+TL008             2       2                             0
+TL009             2       2                             0
+TL010             1       1                             0
+TL011             2       2                             0
+TL012             1       1                             0
+total            18      18                             0
+```
+
+I wrote that corpus together with the rules, so it guards against regressions and says nothing
+about accuracy on code I have not seen.
+
+For a check on someone else's code, I ran it over all 93 Python files of
+[pytorch/examples](https://github.com/pytorch/examples) at commit `acc295d` and read every
+finding:
+
+```console
+git clone https://github.com/pytorch/examples && cd examples
+git checkout acc295dc7b90714f1bf47f06004fc19a7fe235c4
+trainlint check .
+```
+
+It reports 21 findings. By my reading, 19 are real: 15 scripts never set a seed (one declares a
+`--seed` option and never uses it), 3 tensor-parallel scripts never clear gradients, and one script
+resumes from a snapshot without saving optimizer state. Two more are correct as written but
+harmless, because those models have no dropout or batch norm to leave in training mode. The first
+run also flagged three distributed-autograd scripts for missing `zero_grad()`, which was a false
+positive that is now fixed. This is one repository of tutorial code and I did the reviewing, so
+treat it as a spot check. It tells you nothing about what the rules miss.
+
+## Compared with other tools
+
+| Tool | What it covers | Where trainlint differs |
+| --- | --- | --- |
+| ruff, pylint, flake8 | General Python style and bugs | None of their rules know about splits, eval mode or metrics. Use them alongside trainlint. |
+| [TorchFix](https://github.com/meta-pytorch/torchfix) | Removed and deprecated PyTorch functions, `requires_grad` typos, unsafe `torch.load` | Its README lists no rules for training logic. trainlint does not check API deprecations. |
+| [mllint](https://github.com/bvobart/mllint) | Repository-level quality of ML projects: version control, dependencies, code quality | It looks at project hygiene. trainlint reads training and evaluation code. |
+| [Deepchecks](https://github.com/deepchecks/deepchecks) | Checks that run on real datasets and models, including drift | Deepchecks needs your data and model loaded. trainlint reads source only and runs nothing. |
+| General skill collections | Engineering process for coding agents | They are not specific to ML. trainlint's skills target training and evaluation code and come with a checker. |
+
+## Roadmap
+
+- Jupyter notebook support
+- SARIF output for GitHub code scanning
+- Settings in `pyproject.toml`
+- More rules: training loaders without shuffling, `retain_graph` in a loop, normalization on a
+  whole DataFrame before the split
+- A benchmark that measures how often an agent writes these bugs with and without the skills
+  (needs an API key, so it will be optional)
+
+Issues labeled [good first issue](https://github.com/Satvik-jain/trainlint/labels/good%20first%20issue)
+are small and have a clear definition of done.
+
+## Contributing
+
+False positives and missed bugs are the most useful reports. See
+[CONTRIBUTING.md](CONTRIBUTING.md) for setup and for how to add a rule.
+
+```console
+uv sync
+uv run pytest
+uv run ruff check . && uv run ruff format --check .
+uv run python -m bench
+```
+
+## License
+
+MIT
+
+If trainlint saved you an afternoon, a star on the repository helps other people find it.