[brettheap]

Lane: openreposhape-3 (openRepoShape-3)
Claim: #563 (comment)

Refs #563. The workflow PR and the docs PR follow this one, so it does not close the issue.

Stacked on #587; lands after it. The base is l3-564-carry-to-main (#587's branch, head 5191f0eb), because the package re-states the v1.0.1 launcher that #587 carries into main.

What

The first of the lane acts in #563's design (issuecomment-6191672077): the openreposhape Python package. The launcher's and install.sh's symlink refusals are separate small commits, so they merge on their own beside #585 (#596), which edits those two test files' skip conditions and, since #597 removed the macOS and Windows jobs, adds a small Windows job of its own, tests-git-bash, for those two files only. No existing test, skip condition or job step changes here. The four commits below are followed by five fix commits from the first CI round (SonarCloud findings, and the package job's pipx step); the head after that round was d53080ae.

• ba832315 openRepoShape: --install refuses a target that is a symbolic link (pipx places ~/.local/bin/openRepoShape as a link into the package's environment, and cp writes through a link), dangling links included. It places and records nothing, and names both ways out: pipx upgrade openreposhape, or remove the link. This commit is one loop in install_commands plus one new test after test_install_refuses_to_also_scaffold. No existing test or skip condition changes.
• 8cae45cb install.sh: refuses the same target itself before any request, because the release it installs may carry a launcher older than that check. Its header's exit-2 list names the case. One new test in tests/test_install_sh.py.
• 604fbda9 the package:
  • pyproject.toml (hatchling 1.27.0, the newest that runs on Python 3.9; requires-python = ">=3.9"; dynamic = ["version"] read from the stamp; dependencies = ["git-filter-repo==2.47.0"]; five console scripts).
  • src/openreposhape/: _release.py (the stamp: committed as 0.0.0 and unstamped), standard.py (the clone) and cli.py (the commands), all standard-library only and all in SHIPPED.
• 4759eea6 .github/workflows/tests.yml: a new package job, described under Tests. The pytest jobs do not change. Since ci: run the test suite on Linux only #597, main has only one, the Linux tests job.
• Fixes from the first round: ab8bfe34 (launcher test) and bfd927a9 (install.sh and its test) split two composite assertions and use [[. cb910c41 takes the argument loop through one small function per kind of argument, names repeated literals once, and has the tests change sys.path only through monkeypatch; no behaviour changes. 9b0a29e7 and d53080ae run pipx 1.11.1 on the job's Python 3.9, installed from wheels only, because the runner's pipx shares a pip that no longer runs on 3.9.

As designed

• Thin. The wheel holds openreposhape/ plus the tree's setup-project.py byte for byte, beside the modules, so openRepoShape --preflight runs with no git.
• The clone. Every other run uses a full clone of this repository at the stamped 40-hex commit. The clone is made once per commit in the per-user cache ($XDG_CACHE_HOME, ~/Library/Caches, %LOCALAPPDATA%), under openreposhape/checkouts/<commit>, and kept. Before each run with no override, the clone must be at the commit, carry no change (untracked files count), and digest to the stamped value. The clone's own scripts/repo_shape.py computes the digest (sorted-ls-tree-r-v1). The clone is placed whole, so a failed clone leaves nothing behind. A clone that fails a check is refused and left as it is, never repaired.
• The ref. $OPENREPOSHAPE_REF overrides the commit for one run, and no digest is checked then. $OPENREPOSHAPE_REPO names a fork or a mirror. An unstamped build follows main and says so in --version. --version prints the version, tag, commit, digest and clone path, and clones nothing.
• openRepoShape re-states the launcher's <Project>, --preflight, --doctor, --help and --version, and these of its rules:
  • the argument routing, one mode per run, and --org from the flag, else $OPENREPOSHAPE_ORG, else a prompt;
  • the usage text, the ref checks (the launcher's order and words), and --doctor's run, with its kept-clone line (on stderr under --json).
  • A scaffold hands over to the clone's setup-project.py as setup.sh does: --into the typed-in directory first, and OPENREPOSHAPE_SELF_BOOTSTRAP=1.
  • The packaged --install refuses, naming pipx upgrade openreposhape.
• scaffold-project, adopt-project, update-shape, shape-doctor each run that file from the clone, with every argument unchanged, in the directory typed in, and return its exit status.
• git-filter-repo. Every child gets sysconfig.get_path("scripts") first on PATH whenever a git-filter-repo is there. Otherwise PATH is untouched. No tool file changes.

Where it differs from the letter of the design, and why

1. src/openreposhape/, not openreposhape/ at the root. The root already holds openRepoShape, the bash launcher. A directory named openreposhape beside it is the same name on a case-insensitive disk, which is the default on macOS and Windows, so a clone there cannot hold both. The wheel still holds openreposhape/, so the package name, import name and commands are as designed. I could not show this live, because this workstation has no case-insensitive filesystem I can mount.
2. Beside a checkout, the package uses that checkout, as the launcher's --doctor does. This is the case of an editable install, or the test suite running src/. It is what lets the launcher's cases run offline against the working tree.
3. A typed --shape-ref on a scaffold run is that run's ref, over the stamp. In the launcher, setup.sh cloned at it. Passed through alone, it would be ignored silently, because setup-project.py reads it only in self-bootstrap mode.

Why

Brett Heap approved the design at 2026-10-15T13:32:51Z: "approve all six for #563". Constraint 1 holds. "Clone it and run it" still needs only python3 and git. The tools gain no dependency, and the package is a second way in, never a requirement.

Tests

• The launcher's cases drive the Python command. tests/test_openreposhape_package.py imports 13 of the launcher's cases from tests/test_openreposhape_command.py, unchanged. An autouse fixture in the new module points that module's run_cmd and USAGE_LINES at the Python command, for the tests the new module collects. Since ci: run the test suite on Linux only #597, CI runs the suite on Linux only, so these cases run there, and no CI job runs this module on Windows. Five launcher cases are re-stated there for the Python command: the four that run the bash file itself (from stdin or by path), and the description-column case, because the Python command's first usage line differs.
• The package's own cases cover:
  • the stamp and pyproject.toml;
  • --version (stamped, unstamped, beside a checkout, and with an override);
  • the cache: cloned once, reused with the repository gone, and nothing left after a failed clone;
  • the refusals of a changed clone, a clone at another commit, and a wrong digest;
  • $OPENREPOSHAPE_REF, and main followed across a new commit;
  • the four passthroughs (an empty argument, -- and the exit status included);
  • the scaffold hand-over and a typed --shape-ref;
  • --install's refusal, and --preflight with no git on PATH;
  • the PATH for children, both in process and through a real venv whose scripts directory gains a git-filter-repo.
• The package CI job, on ubuntu with no pytest in it, because building the wheel needs its backend and installing it needs git-filter-repo, both from PyPI, and the pytest suite stays network-free (the workflow header says which job is the exception). On Python 3.9, it:
  • builds the wheel twice at the commit's time and cmps the two;
  • checks every wheel file except its metadata against git show HEAD:;
  • installs the wheel with pipx into the job's own directories;
  • runs --version, --help, --install (refused), --preflight, and --doctor --json against this commit served from a bare repository on disk;
  • requires the doctor's machine row to find the git-filter-repo pipx installed beside the package, with none on PATH;
  • runs the four tools' --help.
• Locally, at 4759eea6, and again for the round-2 changes, with a stub gh first on PATH, GH_TOKEN=bogus and an empty GH_CONFIG_DIR:
  • 241 passed in test_openreposhape_package.py, test_openreposhape_command.py, test_install_sh.py and test_repo_hygiene.py;
  • 94 passed in test_repo_hygiene.py and test_cli_reference.py, and 28 in test_cli_reference.py and test_gh_guard.py;
  • 19 passed in the launcher-driven doctor tests of test_shape_doctor.py.
• The package module, two more ways: 54 passed with Path.write_text writing CRLF (a throwaway plugin, not committed), and 53 passed plus 1 skipped (the tomllib check) on Python 3.9.25.
• The CI job's steps, rehearsed locally: they pass, and actionlint 1.7.12 passes over the workflow, with shellcheck.
• By hand, from a git archive of HEAD with a real stamp (commit and digest), installed with pipx:
  • the digest check passed against the clone's own repo_shape.py;
  • a scaffold of Atlas (--local-remote-dir) pinned that commit and digest in project.yaml and contracts/shape-pin.yaml.
• pipx 1.17.12 on Linux exposes the capitalised openRepoShape and does not expose git-filter-repo. A cache hit took 0.86 s, against 5.2 s for the first clone.

Review (RV590) and fix round

The independent review found no MUST and accepted the four departures above. This round answers its six SHOULDs and its nits. The changes are in 66bb2c7e and 6179bd78, with the comment nit in 22ade8e3. They sit on a merge of the base at 25b2aeed (a20c469a, no conflicts).

• A bug, found by a test the review asked for. With $OPENREPOSHAPE_REF naming an annotated tag, remote_commit took the tag object for the commit. git ls-remote -- <url> <ref> prints a tag's peeled line (refs/tags/<ref>^{}) only when a pattern names that line too. The clone then checked out the tag's commit, and verify refused the run because the clone was not at the commit asked for. remote_commit now asks for <ref> and <ref>^{}. A ref here holds only letters, digits and ._/-, so neither pattern is a wildcard. Checked with protocol v0 and v2.
• The same bug, by a tag's full name (re-check of 6179bd78, fixed in d090f0c5). OPENREPOSHAPE_REF=refs/tags/<annotated tag> still resolved to the tag object, because the lookup never tried <ref>^{}, so every such run was refused. The lookup now tries <ref>^{} before <ref>, and the tag test gained the refs/tags/annotated case, which failed before the fix.
• SHOULD 1, the clone is full. test_the_clone_is_full clones through a file:// URL, because git ignores --depth when it clones from a plain path. It checks that --is-shallow-repository prints false and that every commit of the synthetic standard is in the clone.
• SHOULD 2, a bare run asks nothing of git. The bare-run test now runs DETACHED, where the standard is one git call away. The logging git stands first on PATH, and the repository does not exist.
• SHOULD 3, the test runner. PythonCommand.detached now points $OPENREPOSHAPE_REPO at a path that does not exist, as __call__ already did, and __call__ runs in the test's own directory. The two tests that read the default repository's name set the variable to empty, which counts as unset (${VAR:-default}), and a comment says so.
• SHOULD 4, the usage line. It now reads "run from a checkout's src/ (an editable install), it uses that checkout as it stands."
• SHOULD 5, what a killed run leaves. Before a clone is made, sweep_incoming removes any .incoming-* directory that nothing has touched for an hour. It never follows a link, and it skips a directory it cannot remove instead of refusing the run. A test plants one directory two hours old, which is removed, and one fresh directory, which is kept.
• SHOULD 6, the reproducibility claim. The pyproject.toml comment now says only what the package job shows: two builds on one runner, at the commit's time, compared.
• Nits:
  • A tag wins over a branch of the same name, tested with a lightweight tag and with an annotated one.
  • A tool killed by SIGTERM exits 143 (POSIX only). The stub sends itself $STUB_SIGNAL when the variable is set.
  • A new hygiene test, test_the_package_is_pure_ascii, holds every module of src/openreposhape/ to pure ASCII. The existing test is unchanged.
  • The openRepoShape comment over --install's record was reworded. Only that comment changed.
• Each new test fails when its target is broken on purpose:
  • --depth 2 on the clone: shallow is true;
  • the sweep removed: the old directory stays;
  • branches looked up before tags: the wrong ref runs;
  • the peeled pattern dropped: the annotated tag's run is refused;
  • the raw status returned: the exit status is 241.
  • The sources were restored afterwards, with matching sha256.
• Locally at 6179bd78, with the stub gh, GH_TOKEN=bogus and an empty GH_CONFIG_DIR:
  • 250 passed across four files: 58 in the package module, 68 for the launcher, 47 for install.sh and 77 for hygiene;
  • 28 passed in test_cli_reference.py and test_gh_guard.py;
  • 58 passed in the package module with CRLF emulation;
  • 57 passed and 1 skipped on Python 3.9.25.
• Shared files this round: only openRepoShape (the one comment, about line 409). install.sh, tests/test_openreposhape_command.py, tests/test_install_sh.py and the workflow are unchanged.
• CI at d53080ae, before this round: tests (Linux), tests-macos and package passed. The new push superseded tests-windows. ci: run the test suite on Linux only #597 has since removed both the macOS and the Windows jobs.

Verified on CI so far, and still open

• The package job passed at 9b0a29e7. The ubuntu image ships pipx (1.16.7), and actions/setup-python still offers 3.9 (3.9.25). The two builds are one file (b5fdff75…), and every module plus setup-project.py equals its blob at HEAD. pipx 1.11.1 installs the wheel on 3.9.25. The doctor's machine row finds …/pipx/home/venvs/openreposhape/bin/git-filter-repo, with none on PATH.
• The first run failed at the pipx step: the runner's pipx with --python loads its own Python 3.12 pip into the 3.9 environment (TypeError: dataclass() got an unexpected keyword argument 'slots'). Fixed in 9b0a29e7.
• SonarCloud: one finding fails the gate, pythonsecurity:S8705 in cli.run_child. It is the passthrough itself: the person's arguments go to the standard's own script, unchanged, with no shell. The lane rules on it; this PR does not mark it. Since then, lane 1 has marked it, and S8565 on pyproject.toml, false positive on Brett Heap's word, and the gate read OK at 995e1380.
• Still open: the tests and tests-git-bash runs at 4410da00. CI is Linux-only since ci: run the test suite on Linux only #597. pipx with the capitalised command on macOS and Windows, and git for Windows finding a git-filter-repo.exe in a pipx environment's Scripts, are for the release workflow's install check (the next PR), because this job is ubuntu only, as the plan says.

Merge of #587's squash and main

#587 landed as 22eaceed, and this PR now targets main. The branch took #587's final head first, then main, and the head is now 3c208e95.

• 06dad85a merges install.sh and openRepoShape --install, carried into main from v1.0.1 (#577): the one line curl -fsSL https://github.com/opensoft/openRepoShape/releases/latest/download/install.sh | bash installs the newest release's command pinned to that release, installs git-filter-repo with pipx where it can, and ends with the preflight (#564) #587's final head (3d8fa05e).
  • It had one conflict, in .github/workflows/tests.yml. Main's file is kept whole: ci: run the test suite on Linux only #597's Linux-only header and the removal of the macOS and Windows jobs, and .github/workflows/tests.yml: a push run on main is grouped by its commit, so the next landing no longer cancels it, and every landing gets its own Linux result (#592) #593's concurrency: group with its comment.
  • This PR adds only its header paragraph and the package job, which is byte for byte as at d090f0c5.
  • tests/test_repo_hygiene.py merged cleanly. Its diff from 3d8fa05e is exactly this PR's lines.
• 3c208e95 merges origin/main (22eaceed).
  • 22eaceed's tree equals git merge-tree --write-tree 2b6dbb78 3d8fa05e. Main differs from 3d8fa05e only in adopt-project.py: where git checks a symlink out as a plain file, check's finding for a link kept above a path the shape writes says the clone holds it as a file, so scripts -> tools/deep no longer says where the scripts land and sends the person to change where it points, on to adopt-symlinks-off (#531) #584's two files, which merged cleanly.
  • install.sh, openRepoShape and tests/test_install_sh.py conflicted as add/add, because the merge base (62ef256) has none of them.
  • Each of those files on main has the same blob as at 3d8fa05e. This PR's side is that blob plus this PR's own hunks only, so this PR's side was kept.
• Checks:
  • actionlint 1.7.12 passes over tests.yml and the two template workflows.
  • With the stub gh, 326 passed: test_openreposhape_package.py 58, test_install_sh.py 47, test_openreposhape_command.py 68, test_repo_hygiene.py 77 and test_readme_sections.py 76.
  • test_workflow_concurrency.py 5 passed, and the package module passed 58 with CRLF emulation.
  • git merge-tree --write-tree origin/main HEAD is clean and equals HEAD's tree (b94a002d).
• 1c8a7085 changes comments and a docstring only. The package job's comment and the package tests' docstring now describe Linux-only CI.
• Lane 5's four bash_path changes (tests: a tests-git-bash job runs the launcher's and install.sh's suites in Git for Windows' own bash, so README and the handbook say Git Bash runs the one-line install, tested in CI but for a typed yes, which it cannot take there (#585) #596) still pass git apply --check on this tree. They are not applied, because bash_path is not on main yet. They are applied in 4410da00; see below.

Merge of #596 and main

#596 (#585, Git Bash on Windows) landed as bf7b19ba, squashed from 52524fe3. This branch took main at 4cf5fb07 in cfbe59d5, and the head is now 4410da00.

• One conflict, .github/workflows/tests.yml. Main's file is kept whole: ci: run the test suite on Linux only #597's Linux-only tests job, .github/workflows/tests.yml: a push run on main is grouped by its commit, so the next landing no longer cancels it, and every landing gets its own Linux result (#592) #593's concurrency: group, and tests: a tests-git-bash job runs the launcher's and install.sh's suites in Git for Windows' own bash, so README and the handbook say Git Bash runs the one-line install, tested in CI but for a typed yes, which it cannot take there (#585) #596's header exception and tests-git-bash job (Brett Heap's "Option 1: keep a small Git Bash job for tests: a tests-git-bash job runs the launcher's and install.sh's suites in Git for Windows' own bash, so README and the handbook say Git Bash runs the one-line install, tested in CI but for a typed yes, which it cannot take there (#585) #596"). This PR adds its header paragraph after main's, and the package job at the end, byte for byte as before. The jobs are now tests, tests-git-bash and package.
• tests/test_install_sh.py and tests/test_openreposhape_command.py merged cleanly. Each one's diff from main is exactly this PR's symlink cases, and each one's diff from this PR is exactly tests: a tests-git-bash job runs the launcher's and install.sh's suites in Git for Windows' own bash, so README and the handbook say Git Bash runs the one-line install, tested in CI but for a typed yes, which it cannot take there (#585) #596's lines.
• 4410da00 applies lane 5's four bash_path changes, as lane 5 wrote them, in a commit of their own. tests: a tests-git-bash job runs the launcher's and install.sh's suites in Git for Windows' own bash, so README and the handbook say Git Bash runs the one-line install, tested in CI but for a typed yes, which it cannot take there (#585) #596 landed first, so this PR carries them. They make the two symlink refusal cases (four with their dangling parameter) spell paths as Git Bash does, and strip Windows' \\?\ prefix from a link read back. bash_path is on main, in tests/conftest.py.
• 70955e4d changes a comment only. The package job's header paragraph in tests.yml no longer calls it "the one job that reaches PyPI": tests and tests-git-bash pip-install pytest too. What it says instead is the job's own: it builds the wheel and installs it with pipx, which needs PyPI. actionlint, test_workflow_concurrency.py and test_repo_hygiene.py (82 passed) were re-run.
• Checks on Linux, with the stub gh, GH_TOKEN=bogus and an empty GH_CONFIG_DIR:
  • 351 passed: test_openreposhape_package.py 58, test_install_sh.py 47, test_install_sh_filter_repo.py 18, test_openreposhape_command.py 70, test_repo_hygiene.py 77, test_readme_sections.py 76 and test_workflow_concurrency.py 5;
  • the two tests-git-bash suites gave 117 passed, both plain and with Path.write_text writing CRLF (a throwaway plugin, not committed);
  • the package module gave 58 passed with CRLF;
  • actionlint 1.7.12 exits 0;
  • git merge-tree --write-tree origin/main HEAD is clean and equals HEAD's tree (78125346).

🤖 Generated with Claude Code

Summary by Sourcery

Add a pipx-installable openreposhape package backed by verified standard checkouts, and protect both installers from symlink targets.

New Features:

• Add the openreposhape Python package with openRepoShape and four standard-tool console commands that execute from a verified checkout of the repository standard.
• Add stamped release metadata, cached full checkouts, reference overrides, digest validation, passthrough execution, preflight support, and package-managed git-filter-repo discovery.
• Add CI coverage that builds reproducible wheels, validates their contents, installs them with pipx on Python 3.9, and exercises the packaged commands.

Bug Fixes:

• Prevent openRepoShape --install and install.sh from writing through live or dangling symbolic-link targets, with guidance for resolving pipx-managed installations.

Enhancements:

• Preserve launcher-compatible argument handling, diagnostics, doctor behavior, scaffold handoff, and child exit-status propagation in the packaged commands.
• Sweep stale interrupted clone directories and refuse altered, incomplete, or digest-mismatched cached standards without repairing them.

Build:

• Introduce a Hatchling-based Python package build configured for Python 3.9+, dynamic release versions, bundled setup-project.py, and the pinned git-filter-repo dependency.

CI:

• Add a dedicated Ubuntu package job that builds and compares wheels, verifies source contents, installs with pipx, and runs packaged command smoke checks.

Tests:

• Add comprehensive package tests covering command compatibility, release stamps, checkout caching and verification, reference resolution, passthrough tools, preflight behavior, PATH handling, and signal propagation.
• Add symlink refusal tests for both installers and package hygiene checks for shipped modules and ASCII sources.

[sourcery-ai]

Reviewer's Guide

Introduces the openreposhape package as a second installation path with five console commands that run the repository standard from a cached, clean, commit- and digest-verified checkout, adds reproducible pipx packaging CI and broad behavioral coverage, and hardens both installers against symlink targets.

Sequence diagram for a verified package command run

sequenceDiagram
    participant User
    participant CLI as openRepoShape CLI
    participant Standard as standard.prepare
    participant Git
    participant Cache
    participant Tool as Cloned standard tool
    User->>CLI: Run command
    CLI->>Standard: prepare()
    Standard->>Git: Resolve stamped commit or override ref
    Standard->>Cache: Find or clone checkout
    Cache-->>Standard: Checkout path
    Standard->>Git: Verify HEAD and clean status
    Standard->>Standard: tree_digest()
    Standard-->>CLI: Verified Standard
    CLI->>Tool: run_child(arguments unchanged)
    Tool-->>CLI: Exit status
    CLI-->>User: Return tool status

Loading

Sequence diagram for symlink-safe installation refusal

sequenceDiagram
    participant User
    participant Installer
    participant Target as openRepoShape target
    User->>Installer: Run --install
    Installer->>Target: Check symbolic link
    alt Target is symbolic link
        Installer-->>User: Refuse before network or copying
        Installer-->>User: Suggest pipx upgrade or remove link
    else Target is not symbolic link
        Installer->>Installer: Fetch and install launcher
    end

Loading

File-Level Changes

Change	Details	Files
Adds a distributable Python package that reimplements the launcher and exposes four pass-through tool commands while executing standard files from a verified repository checkout.	Defines a Python 3.9-compatible Hatchling build with a stamped dynamic version, pinned git-filter-repo dependency, five console scripts, and bundled setup-project.py. Implements stamp handling, checkout selection, cache reuse, repository/ref resolution, atomic cloning, clean-tree and commit verification, and stamped tree-digest validation. Recreates launcher routing, usage, preflight, doctor, scaffold hand-off, environment handling, and exit-status behavior; passes tool arguments unchanged and adds the package scripts directory to child PATH when needed. Adds extensive offline package tests, shared launcher behavior coverage, package hygiene checks, and a Python 3.9 CI packaging job that validates reproducible wheels, wheel contents, pipx installation, and command execution.	pyproject.toml src/openreposhape/__init__.py src/openreposhape/__main__.py src/openreposhape/_release.py src/openreposhape/cli.py src/openreposhape/standard.py tests/test_openreposhape_package.py tests/test_repo_hygiene.py .github/workflows/tests.yml
Prevents installers from overwriting a pipx-managed command through a symbolic link.	Makes packaged --install reject live and dangling openRepoShape symlinks before writing or recording anything, with pipx upgrade and uninstall/removal guidance. Makes install.sh perform the same symlink check before network access so older downloaded launchers cannot bypass it. Adds live and dangling symlink regression tests for both installation paths.	openRepoShape install.sh tests/test_openreposhape_command.py tests/test_install_sh.py

Possibly linked issues

• Publish openRepoShape to TestPyPI and PyPI as openreposhape, so one pipx install puts the tools and git-filter-repo on any machine (design first) #563: The PR is the package implementation specified by issue Publish openRepoShape to TestPyPI and PyPI as openreposhape, so one pipx install puts the tools and git-filter-repo on any machine (design first) #563; release workflow and documentation are intentionally deferred.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[sourcery-ai]

Sorry @brettheap, you've used your own review budget of 250,000 diff characters for the last 7 days.

You can request another review in 1 day and 15 hours by commenting @sourcery-ai review. Upgrade to get a review now.