acumen 0.0.1.dev0__tar.gz → 0.0.2__tar.gz

acumen-0.0.2/.codecov.yaml

@@ -0,0 +1,17 @@
+# Based on pydata/xarray
+codecov:
+    require_ci_to_pass: no
+
+coverage:
+    status:
+        project:
+            default:
+                # Require 1% coverage, i.e., always succeed
+                target: 1
+        patch: false
+        changes: false
+
+comment:
+    layout: diff, flags, files
+    behavior: once
+    require_base: no

acumen-0.0.2/.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,93 @@
+name: Bug report
+description: Report something that is broken or incorrect
+labels: bug
+body:
+    - type: markdown
+      attributes:
+          value: |
+              **Note**: Please read [this guide](https://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports)
+              detailing how to provide the necessary information for us to reproduce your bug. In brief:
+                * Please provide exact steps how to reproduce the bug in a clean Python environment.
+                * In case it's not clear what's causing this bug, please provide the data or the data generation procedure.
+                * Replicate problems on public datasets or share data subsets when full sharing isn't possible.
+
+    - type: textarea
+      id: report
+      attributes:
+          label: Report
+          description: A clear and concise description of what the bug is.
+      validations:
+          required: true
+
+    - type: textarea
+      id: versions
+      attributes:
+          label: Versions
+          description: |
+              Which version of packages.
+
+              Please install `session-info2`, run the following command in a notebook,
+              click the “Copy as Markdown” button, then paste the results into the text box below.
+
+              ```python
+              In[1]: import session_info2; session_info2.session_info(dependencies=True)
+              ```
+
+              Alternatively, run this in a console:
+
+              ```python
+              >>> import session_info2; print(session_info2.session_info(dependencies=True)._repr_mimebundle_()["text/markdown"])
+              ```
+          render: python
+          placeholder: |
+              anndata	0.11.3
+              ----	----
+              charset-normalizer	3.4.1
+              coverage	7.7.0
+              psutil	7.0.0
+              dask	2024.7.1
+              jaraco.context	5.3.0
+              numcodecs	0.15.1
+              jaraco.functools	4.0.1
+              Jinja2	3.1.6
+              sphinxcontrib-jsmath	1.0.1
+              sphinxcontrib-htmlhelp	2.1.0
+              toolz	1.0.0
+              session-info2	0.1.2
+              PyYAML	6.0.2
+              llvmlite	0.44.0
+              scipy	1.15.2
+              pandas	2.2.3
+              sphinxcontrib-devhelp	2.0.0
+              h5py	3.13.0
+              tblib	3.0.0
+              setuptools-scm	8.2.0
+              more-itertools	10.3.0
+              msgpack	1.1.0
+              sparse	0.15.5
+              wrapt	1.17.2
+              jaraco.collections	5.1.0
+              numba	0.61.0
+              pyarrow	19.0.1
+              pytz	2025.1
+              MarkupSafe	3.0.2
+              crc32c	2.7.1
+              sphinxcontrib-qthelp	2.0.0
+              sphinxcontrib-serializinghtml	2.0.0
+              zarr	2.18.4
+              asciitree	0.3.3
+              six	1.17.0
+              sphinxcontrib-applehelp	2.0.0
+              numpy	2.1.3
+              cloudpickle	3.1.1
+              sphinxcontrib-bibtex	2.6.3
+              natsort	8.4.0
+              jaraco.text	3.12.1
+              setuptools	76.1.0
+              Deprecated	1.2.18
+              packaging	24.2
+              python-dateutil	2.9.0.post0
+              ----	----
+              Python	3.13.2 | packaged by conda-forge | (main, Feb 17 2025, 14:10:22) [GCC 13.3.0]
+              OS	Linux-6.11.0-109019-tuxedo-x86_64-with-glibc2.39
+              Updated	2025-03-18 15:47

acumen-0.0.2/.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,5 @@
+blank_issues_enabled: false
+contact_links:
+    - name: Scverse Community Forum
+      url: https://discourse.scverse.org/
+      about: If you have questions about “How to do X”, please ask them here.

acumen-0.0.2/.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,11 @@
+name: Feature request
+description: Propose a new feature for acumen
+labels: enhancement
+body:
+    - type: textarea
+      id: description
+      attributes:
+          label: Description of feature
+          description: Please describe your suggestion for a new feature. It might help to describe a problem or use case, plus any alternatives that you have considered.
+      validations:
+          required: true

acumen-0.0.2/.github/workflows/build.yaml

@@ -0,0 +1,26 @@
+name: Check Build
+
+on:
+    push:
+        branches: [main]
+    pull_request:
+        branches: [main]
+
+concurrency:
+    group: ${{ github.workflow }}-${{ github.ref }}
+    cancel-in-progress: true
+
+jobs:
+    package:
+        runs-on: ubuntu-latest
+        steps:
+            - uses: actions/checkout@v6
+              with:
+                  filter: blob:none
+                  fetch-depth: 0
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+            - name: Build package
+              run: uv build
+            - name: Check package
+              run: uvx twine check --strict dist/*.whl

acumen-0.0.2/.github/workflows/release.yaml

@@ -0,0 +1,27 @@
+name: Release
+
+on:
+    release:
+        types: [published]
+
+# Use "trusted publishing", see https://docs.pypi.org/trusted-publishers/
+jobs:
+    release:
+        name: Upload release to PyPI
+        runs-on: ubuntu-latest
+        environment:
+            name: pypi
+            url: https://pypi.org/p/acumen
+        permissions:
+            id-token: write # IMPORTANT: this permission is mandatory for trusted publishing
+        steps:
+            - uses: actions/checkout@v6
+              with:
+                  filter: blob:none
+                  fetch-depth: 0
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+            - name: Build package
+              run: uv build
+            - name: Publish package distributions to PyPI
+              uses: pypa/gh-action-pypi-publish@release/v1

acumen-0.0.2/.github/workflows/test.yaml

@@ -0,0 +1,113 @@
+name: Test
+
+on:
+    push:
+        branches: [main]
+    pull_request:
+        branches: [main]
+    schedule:
+        - cron: "0 5 1,15 * *"
+
+concurrency:
+    group: ${{ github.workflow }}-${{ github.ref }}
+    cancel-in-progress: true
+
+jobs:
+    # Get the test environment from hatch as defined in pyproject.toml.
+    # This ensures that the pyproject.toml is the single point of truth for test definitions and the same tests are
+    # run locally and on continuous integration.
+    # Check [[tool.hatch.envs.hatch-test.matrix]] in pyproject.toml and https://hatch.pypa.io/latest/environment/ for
+    # more details.
+    get-environments:
+        runs-on: ubuntu-slim
+        outputs:
+            envs: ${{ steps.get-envs.outputs.envs }}
+        steps:
+            - uses: actions/checkout@v6
+              with:
+                  filter: blob:none
+                  fetch-depth: 0
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+            - name: Get test environments
+              id: get-envs
+              run: |
+                  ENVS_JSON=$(uvx hatch env show --json | jq -c 'to_entries
+                    | map(
+                        select(.key | startswith("hatch-test"))
+                        | {
+                            name: .key,
+                            label: (if (.key | contains("pre")) then .key + " (PRE-RELEASE DEPENDENCIES)" else .key end),
+                            python: .value.python
+                          }
+                      )')
+                  echo "envs=${ENVS_JSON}" | tee $GITHUB_OUTPUT
+
+    # Run tests through hatch. Spawns a separate runner for each environment defined in the hatch matrix obtained above.
+    test:
+        needs: get-environments
+        permissions:
+            id-token: write # for codecov OIDC
+            contents: read
+
+        strategy:
+            fail-fast: false
+            matrix:
+                os: [ubuntu-latest]
+                env: ${{ fromJSON(needs.get-environments.outputs.envs) }}
+
+        name: ${{ matrix.env.label }}
+        runs-on: ${{ matrix.os }}
+        continue-on-error: ${{ contains(matrix.env.name, 'pre') }} # make "all-green" pass even if pre-release job fails
+
+        steps:
+            - uses: actions/checkout@v6
+              with:
+                  filter: blob:none
+                  fetch-depth: 0
+            - name: Set up Node.js
+              uses: actions/setup-node@v6
+              with:
+                  node-version: 22
+            - name: Install uv
+              uses: astral-sh/setup-uv@v7
+              with:
+                  python-version: ${{ matrix.env.python }}
+            - name: Install agent CLIs
+              run: npm install --global @anthropic-ai/claude-code @openai/codex
+            - name: create hatch environment
+              run: uvx hatch env create ${{ matrix.env.name }}
+            - name: list all all installed package versions
+              run: uvx hatch run ${{ matrix.env.name }}:uv pip list
+            - name: run tests using hatch
+              env:
+                  MPLBACKEND: agg
+                  PLATFORM: ${{ matrix.os }}
+                  DISPLAY: :42
+              run: uvx hatch run ${{ matrix.env.name }}:run-cov -v --color=yes -n auto
+            - name: generate coverage report
+              run: |
+                  # See https://coverage.readthedocs.io/page/config.html#run-patch
+                  test -f .coverage || uvx hatch run ${{ matrix.env.name }}:cov-combine
+                  uvx hatch run ${{ matrix.env.name }}:cov-report   # report visibly
+                  uvx hatch run ${{ matrix.env.name }}:coverage xml # create report for upload
+            - name: Upload coverage
+              uses: codecov/codecov-action@v6
+              with:
+                  # TODO: set back to true once scverse/acumen is onboarded on Codecov
+                  fail_ci_if_error: false
+                  use_oidc: true
+
+    # Check that all tests defined above pass. This makes it easy to set a single "required" test in branch
+    # protection instead of having to update it frequently. See https://github.com/re-actors/alls-green#why.
+    check:
+        name: Tests pass in all hatch environments
+        if: always()
+        needs:
+            - get-environments
+            - test
+        runs-on: ubuntu-latest
+        steps:
+            - uses: re-actors/alls-green@release/v1
+              with:
+                  jobs: ${{ toJSON(needs) }}

{acumen-0.0.1.dev0 → acumen-0.0.2}/.gitignore

@@ -11,6 +11,10 @@ __pycache__/
 # Distribution / packaging
 /dist/
 
+# uv's resolved dependency set. acumen is a library — consumers resolve their own tree, and
+# CI resolves fresh through hatch — so a committed lock would pin nothing that is checked.
+/uv.lock
+
 # Tests and coverage
 /data/
 /node_modules/

acumen-0.0.2/.pre-commit-config.yaml

@@ -0,0 +1,38 @@
+fail_fast: false
+default_language_version:
+    python: python3
+default_stages:
+    - pre-commit
+    - pre-push
+minimum_pre_commit_version: 2.16.0
+repos:
+    - repo: https://github.com/biomejs/pre-commit
+      rev: v2.6.4
+      hooks:
+          - id: biome-format
+            exclude: ^\.cruft\.json$ # inconsistent indentation with cruft - file never to be modified manually.
+    - repo: https://github.com/tox-dev/pyproject-fmt
+      rev: v2.30.6
+      hooks:
+          - id: pyproject-fmt
+    - repo: https://github.com/astral-sh/ruff-pre-commit
+      rev: v0.17.4
+      hooks:
+          - id: ruff-check
+            types_or: [python, pyi, jupyter]
+            args: [--fix, --exit-non-zero-on-fix]
+          - id: ruff-format
+            types_or: [python, pyi, jupyter]
+    - repo: https://github.com/pre-commit/pre-commit-hooks
+      rev: v6.0.0
+      hooks:
+          - id: detect-private-key
+          - id: check-ast
+          - id: end-of-file-fixer
+          - id: mixed-line-ending
+            args: [--fix=lf]
+          - id: trailing-whitespace
+          - id: check-case-conflict
+          # Check that there are no merge conflicts (could be generated by template sync)
+          - id: check-merge-conflict
+            args: [--assume-in-merge]

acumen-0.0.2/.readthedocs.yaml

@@ -0,0 +1,16 @@
+# https://docs.readthedocs.io/page/config-file/v2.html
+version: 2
+build:
+    os: ubuntu-24.04
+    tools:
+        python: "3.14"
+        nodejs: latest
+    jobs:
+        create_environment:
+            - asdf plugin add uv
+            - asdf install uv latest
+            - asdf global uv latest
+        build:
+            html:
+                - uvx hatch run docs:build
+                - mv docs/_build $READTHEDOCS_OUTPUT

{acumen-0.0.1.dev0 → acumen-0.0.2}/.vscode/settings.json

@@ -15,4 +15,5 @@
     "python.analysis.typeCheckingMode": "basic",
     "python.testing.pytestEnabled": true,
     "python.testing.pytestArgs": ["-vv", "--color=yes"],
+    "cursorpyright.analysis.typeCheckingMode": "basic",
 }

acumen-0.0.2/CHANGELOG.md

@@ -0,0 +1,74 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog][],
+and this project adheres to [Semantic Versioning][].
+
+[keep a changelog]: https://keepachangelog.com/
+[semantic versioning]: https://semver.org/
+
+## [0.0.2]
+
+### Added
+
+- Added `acumen fit --ablation {no-optimize,no-wiki}` to run ablation arms. A report without a `valid` split now uses `test` instead
+- Added `acumen prepare` to build task datasets once into `cache/`, and a `data:` key in `tasks.yaml` that links them into each run
+
+### Changed
+
+- Restructured the docs, with a scverse-style README, a Getting started tutorial and one API page per `acumen` subcommand
+- `--agent codex` now installs shipped skills into `~/.agents/skills`, where current Codex looks for them
+- The report now tests success rate and cost separately, with a Benjamini–Hochberg FDR correction per axis
+- Updated the GPT colour ramp in the report so `astra` is the darkest and the other models lighten in even steps
+- Merged the per-command model keys (`draft_model`, `improve_model`, `tasks_model`, `ship_model`, `check_model`) into a single `meta_model` key
+- Transcripts from every provider are now rendered through one shared trajectory model, and each run also writes a `trajectory.json`. `claude-code-log` is no longer a dependency
+- Claude agents now run with the `claude` CLI on `PATH`, falling back to the one bundled in `claude-agent-sdk`
+- `result.json` now records the agent CLI's version as `cli_version`, replacing `sdk_version`
+
+### Fixed
+
+- Fixed agent failures reporting a bare "errored: success" instead of the provider's reason
+- Fixed `acumen ship --force` leaving the old installer's tests broken. The ship agent now ports them and runs them before delivering
+- Fixed `acumen ship` failing to push and open the pull request because the agent could not find the operator's git and `gh` credentials
+
+## [0.0.1dev]
+
+### Added
+
+- `acumen check` now also reviews each task for inconsistencies between its prompt, answer and reproducer. `--no-review` skips this step
+- Added `acumen check` to verify each task's answer by rerunning its reproducer at `tasks/<id>-<split>.py`
+- `acumen tasks` now keeps one reproducer script per split in `tasks/`
+- Added an optional `needs_script` key to `tasks.yaml` for tasks that need no code
+- Claude and Codex models can now run side by side in benchmarks and drive every meta-agent command
+- `cost_usd` is now computed from each run's token counts, so both providers are priced the same way
+- Added `acumen prices` to show and refresh the rate table, and a `prices:` config key to override it
+- Rates are now read from the providers' pricing pages instead of a table shipped with the release
+- `acumen bench` now fails before spending anything when the pricing pages cannot be read
+- `result.json` now records where each run's prices came from and their date
+- The report now flags arms that were priced on different dates
+- Added HTML transcripts for Codex runs
+- `acumen bench` without an arm now benches the baseline and every version in `skills/` in one pass
+
+### Changed
+
+- The report's cost vs. success figure now joins each model's versions with arrows instead of using marker shapes
+- The Pareto frontier in the report is now dashed and runs to the right edge of the panel
+- Both backends are now optional, installed with `acumen[claude]`, `acumen[all]` or just the `codex` CLI for Codex
+- `max_turns` and `max_usd` are now enforced for Codex
+- `acumen bench` now supports `--auth {auto,session,api}` and defaults to the provider's subscription
+- Runs that fail because provider usage or credit ran out are now treated as invalid and retried on resume
+
+### Fixed
+
+- Fixed baseline runs finding a skill that the target package ships inside itself. The target venv is now cleaned of agent guidance before any agent runs
+- Fixed Claude runs being recorded with the wrong result when a background command kept the session alive
+- Increased the Bash timeout for sandboxed commands to 600 seconds, so long downloads are no longer sent to the background
+- Fixed capped or crashed Codex runs recording a cost of `$0.00`
+- Fixed capped runs being failed even when they had already written a correct answer
+- Fixed sandboxed agents being blocked from hosts that targets need, such as EBI, Zenodo and NCBI
+- Fixed runs whose sandbox could not execute anything being graded as wrong answers instead of infrastructure failures
+- `acumen tasks` now overwrites the untouched `tasks.yaml` placeholder from `acumen init` without `--force`
+- Fixed Codex transcripts being rendered as empty pages
+- Removed "Claude" from the drafting and improving prompts, which are also used by Codex
+- Removed `check_auth` and `auth_available`, which ignored Codex credentials

acumen-0.0.2/PKG-INFO

@@ -0,0 +1,161 @@
+Metadata-Version: 2.5
+Name: acumen
+Version: 0.0.2
+Summary: Build, Benchmark, and Optimize Agentic Skills for Python Packages.
+Project-URL: Documentation, https://acumen-skills.readthedocs.io/
+Project-URL: Homepage, https://github.com/scverse/acumen
+Project-URL: Source, https://github.com/scverse/acumen
+Author: Pau Badia i Mompel
+Maintainer-email: Pau Badia i Mompel <paubim@stanford.edu>
+License: MIT License
+        
+        Copyright (c) 2026, Pau Badia i Mompel
+        
+        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.
+License-File: LICENSE
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Programming Language :: Python :: 3.15
+Requires-Python: >=3.12
+Requires-Dist: matplotlib
+Requires-Dist: numpy
+Requires-Dist: pandas
+Requires-Dist: psutil>=5.9
+Requires-Dist: pyyaml
+Requires-Dist: scipy
+Requires-Dist: session-info2
+Provides-Extra: all
+Requires-Dist: claude-agent-sdk>=0.2.126; extra == 'all'
+Provides-Extra: claude
+Requires-Dist: claude-agent-sdk>=0.2.126; extra == 'claude'
+Description-Content-Type: text/markdown
+
+<p align="center">
+  <img src="docs/_static/images/banner.svg" alt="acumen" width="360">
+</p>
+
+<!-- docs-start -->
+
+<p align="center"><b>Build, benchmark, and optimize agentic skills for Python packages.</b></p>
+
+[![Tests][badge-tests]][tests]
+[![Documentation][badge-docs]][documentation]
+
+[![Issues][badge-issues]][issue tracker]
+[![Coverage][badge-coverage]][codecoverage]
+[![Stars][badge-stars]](https://github.com/scverse/acumen/stargazers)
+
+[![PyPI][badge-pypi]][pypi]
+[![Downloads month][badge-mdown]][down]
+[![Downloads all][badge-adown]][down]
+
+[badge-tests]: https://img.shields.io/github/actions/workflow/status/scverse/acumen/test.yaml?branch=main
+[badge-docs]: https://img.shields.io/readthedocs/acumen-skills
+[badge-issues]: https://img.shields.io/github/issues/scverse/acumen
+[badge-coverage]: https://codecov.io/gh/scverse/acumen/branch/main/graph/badge.svg
+[badge-pypi]: https://img.shields.io/pypi/v/acumen.svg
+[badge-mdown]: https://static.pepy.tech/badge/acumen/month
+[badge-adown]: https://static.pepy.tech/badge/acumen
+[badge-stars]: https://img.shields.io/github/stars/scverse/acumen?style=flat&logo=github&color=yellow
+
+`acumen` is a python package to build, benchmark and optimize agentic skills for any Python package.
+An agentic skill is a short set of plain-text instructions that helps a coding agent use a package correctly.
+Given a package and a few tasks, `acumen` trains a skill in the same way one would train a model.
+In every epoch, coding agents attempt the training tasks, and `acumen` learns from the errors they make to rewrite the skill.
+Each new version is scored on held-out validation tasks, training stops once validation no longer improves, and the best version is evaluated one last time on an untouched test set.
+With the final skill, coding agents are more successful at solving tasks that require using the Python package than they are without the skill.
+
+[//]: # "numfocus-fiscal-sponsor-attribution"
+
+`acumen` is part of the scverse® project ([website](https://scverse.org), [governance](https://scverse.org/about/roles)) and is fiscally sponsored by [NumFOCUS](https://numfocus.org/).
+If you like scverse® and want to support our mission, please consider making a tax-deductible [donation](https://numfocus.org/donate-to-scverse) to help the project pay for developer time, professional services, travel, workshops, and a variety of other needs.
+
+<div align="center">
+<a href="https://numfocus.org/project/scverse">
+  <img
+    src="https://raw.githubusercontent.com/numfocus/templates/master/images/numfocus-logo.png"
+    width="200"
+  >
+</a>
+</div>
+
+## Getting started
+
+Please refer to the [documentation][],
+in particular, the [Getting started][getting started] tutorial and the [API documentation][].
+
+## Installation
+
+You need to have Python 3.12 or newer installed on your system.
+If you don't have Python installed, we recommend installing [uv][].
+
+`acumen` runs its agents with Claude, Codex, or both, and you only need to install the backend you use.
+There are several alternative options to install `acumen`:
+
+1. Install the latest stable release from [PyPI][pypi] to run Claude models (requires an Anthropic API key or a `claude` login):
+
+```bash
+pip install acumen[claude]
+```
+
+2. Install the latest stable release from [PyPI][pypi] to run Codex models (requires the `codex` CLI on your `PATH` and a Codex login or API key):
+
+```bash
+pip install acumen
+```
+
+3. Install the latest stable release from [PyPI][pypi] to run both:
+
+```bash
+pip install acumen[all]
+```
+
+4. Install the latest development version:
+
+```bash
+pip install "acumen[all] @ git+https://github.com/scverse/acumen.git@main"
+```
+
+## Release notes
+
+See the [changelog][].
+
+## Contact
+
+For questions and help requests, you can reach out in the [scverse discourse][].
+If you found a bug, please use the [issue tracker][].
+
+## Citation
+
+> t.b.a
+
+[uv]: https://github.com/astral-sh/uv
+[scverse discourse]: https://discourse.scverse.org/
+[scverse]: https://scverse.org/
+[issue tracker]: https://github.com/scverse/acumen/issues
+[tests]: https://github.com/scverse/acumen/actions/workflows/test.yaml
+[documentation]: https://acumen-skills.readthedocs.io/
+[getting started]: https://acumen-skills.readthedocs.io/page/getting_started.html
+[changelog]: https://acumen-skills.readthedocs.io/page/changelog.html
+[api documentation]: https://acumen-skills.readthedocs.io/page/api/index.html
+[pypi]: https://pypi.org/project/acumen
+[down]: https://pepy.tech/project/acumen
+[codecoverage]: https://codecov.io/gh/scverse/acumen