Shell good first issues

The `grit gc --dry-run` command incorrectly reports 'nothing to pack' despite there being loose objects that would be packed. This is inconsistent with the JSON output of the same command, the output of `grit repack --dry-run`, and the actual outcome of `grit gc`.

This issue proposes adding a 1-click deploy option to Heroku, leveraging Docker deployment methods. The contributor notes that a similar feature has already been implemented in Rocket.Chat, suggesting a precedent and potential for reuse of existing patterns.

This issue is a request to add the 'Whistle' model from cactuscompute.com to the project. The user highlights that the model is very effective and lightweight, suggesting it would be a valuable addition.

The issue proposes creating a Pacman repository to host pre-compiled binaries, similar to existing projects like devkitPro and Cpasjuste's PacBrew. This would streamline releases, especially given current challenges with LLVM, and the author is seeking community interest and potential help.

This issue highlights a need to expand the test corpus for EMBA's binary detection rules. The project has created many test files but requires more real-world binary data to ensure comprehensive testing. Contributors are encouraged to run EMBA scans with their firmware and submit the generated corpus files to help complete the test suite.

This issue addresses intermittent failures in the `surface.test.sh` test suite, which are caused by browser timing races in UI journey checks. The fix involves making these checks wait for the specific UI states they assert, rather than relying on fixed delays, to ensure test stability and reliability.

This issue proposes integrating a new 'sweep-only' mode for audit sessions into the 'go-faster' performance plugin. This change will allow 'go-faster' to correctly report on the 'how-you-work' area and update its evaluation logic, even when the session flow store is not yet populated.

Parallel workers are colliding on shared scratchpad script names, leading to incorrect edits and manual reverts. The proposed solution involves naming scratch files with a branch and purpose identifier and preventing workers from running scripts they didn't create, with a potential per-worker scratch subdirectory.

This issue proposes making native Wine Wayland support the default for Wine applications running on Wayland display servers. The author believes this is now stable and potentially more performant than using XWayland, suggesting the removal of experimental flags and the introduction of a disable flag for this new default.

A smoke test for the video digest skill's acquisition module is timing out locally due to reading the entire OS temporary directory. The proposed fix is to isolate the test to a newly created temporary directory with a few known files, or to remove the test if its functionality is already covered by unit tests.

This issue identifies three gaps in the `discipline` plugin's evaluation cases: a lack of tests for fan-out retries, stale fact recall, and the sweep-all fork guard. The suggested fix involves adding specific fixtures and modifying prompts to ensure these functionalities are properly tested and verified.

The `list-promotion-candidates.js` script fails to handle errors when processing a malformed claim inventory, resulting in an uncaught stack trace instead of a user-friendly error message and a non-zero exit code. The fix involves implementing try-catch blocks similar to a sibling script and adding a test case for this scenario.

This issue proposes improvements to the `/user-experience:evaluate` tool to address its failure to detect critical usability problems. The suggested changes include ensuring primary actions are visible without scrolling, identifying ambiguous in-app copy, and verifying the accuracy and relevance of in-product screenshots.

The `nested-agents-check.test.sh` script in the harness-memory plugin is printing a git commit failure message before successfully passing the test. This is due to the fixture helper `commit_all` not checking the result of its commit operation. The proposed solution involves identifying the cause of the fixture commit failure and ensuring the test fails if a fixture commit fails.

This issue proposes updating documentation in `SKILL.md` to reflect current timeout configurations for the Bash tool. The existing text incorrectly claims a hard 10-minute timeout and unlimited background execution, which has been superseded by configurable limits and session-specific time limits for background commands.

A Python test suite for `harness-ops/audit-install-state` fails on Windows when executed from a Git worktree. The issue arises because the absolute path to the test script is being misinterpreted as a module name by the unittest framework. The proposed solution involves adjusting how the Python tests are invoked to ensure they work correctly regardless of the current working directory or whether a worktree is in use.

This issue proposes adding a new evaluation case for the code-tidying batch-simplify feature. The goal is to create a test scenario that specifically excludes task-related tools, thereby exercising and validating the skill's fallback behavior when task tools are not available. This will ensure the code-tidying functionality works correctly in a wider range of conditions.

This issue addresses a bug where temporary directories created by `run-watch.js` are not cleaned up if the process is interrupted by a signal (like Ctrl+C) before it fully starts. The fix involves registering signal handlers to perform the cleanup and exit gracefully, preventing orphaned temporary files.

This issue identifies two bugs in the Python `yaml_subset.py` script related to YAML parsing. The first bug causes trailing comments to be swallowed when a quote appears mid-value, and the second incorrectly reports the line number for duplicate keys within sequence items. The issue requests failing tests to reproduce these bugs before implementing fixes, which are based on solutions already implemented in the JavaScript port.

This issue proposes to refactor the CI process by moving shared library carrier plugins to use changelog fragments. This change aims to reduce version-bump conflicts between parallel pull requests, as evidenced by recent merge issues. The goal is to simplify the CI workflow and improve developer efficiency by preventing redundant merge and CI rounds.

A Python test script on Windows fails because it attempts to execute a `#!/bin/sh` shim directly, which Windows cannot do. This is a platform-specific issue preventing a specific test from running correctly on Windows environments.

This issue identifies a bug in the skill wording for settings-token fallback within the evals plugin. Specifically, a token is used without a fallback, leading to incorrect interpretation when unexpanded. The fix involves updating the wording to include a fallback and removing an unnecessary entry from a baseline file.

This issue addresses missing "Next" sections in documentation for several SDLC skills. The goal is to update specific skill files to include these sections, following a defined workflow chain, to ensure documentation consistency and adherence to established rules. The proposed fix involves adding the "Next" sections to the listed skills and ensuring the "plan" skill's "Next" section accurately reflects its outcomes.

The documentation for humanlayer-skills incorrectly states that the MIT license notice is reproduced below a specific paragraph. In reality, the notice appears earlier in the document, and the text needs to be updated to accurately reflect its location.

This issue requires adding two new rows to the "Implementers" section of the configuration cascade documentation. These rows will document the `source-control.yaml` and `work-items.yaml` files, specifying their current status as hand-edited until relevant setup skills are updated to write them.

This issue proposes to move lane outcome records from GitHub check run annotations to the `output.summary` and `output.text` fields of check runs. This change aims to simplify data retrieval for various consumers (pr-unblock dispatchers, morning brief, babysit scripts) by eliminating the need for extra API calls to fetch annotations. The proposal also addresses how job-level lanes will continue to be handled, acknowledging potential annotation limits.

This issue suggests that the program's usefulness is currently limited by its lack of popularity. The author proposes exploring the activation of a specific URL, https://ur.humaninstrumentalityproject.org, as a potential solution to increase its reach and, consequently, its usefulness. Related censorship issues are also referenced.

This issue requests adding PowerShell commands to the "Verify a release" documentation page for Windows users. The goal is to provide equivalent instructions for calculating file hashes and verifying signatures, mirroring the existing Bash commands. The author also emphasizes testing these commands on a real Windows release before documenting them.

This issue proposes adding a new `--tool-version` command-line option to the `scan-sbom.sh` script. This option will print the version of the BomLens scanner image that the script is configured to use, without performing a scan. This will help users and scripts identify the exact scanner version for debugging and reproducibility.

This issue is a task to add symlinks and icons for various Flathub applications. The list includes several applications that still need this functionality implemented, with one application already marked as complete. It appears to be a recurring task to ensure all Flathub apps have proper desktop integration.