Live good-first-issue & help-wanted feed

Find meaningful ways to contribute

Browse open GitHub issues that maintainers labeled for help. Filter by language, stars, category, and difficulty — then pick something you can finish.

Projects
119,969
Open issues
833,825
Cost
Free
Illustration of a Good first issue card with GitHub, plus Learn new skills, Work on real projects, Make an impact, and Open source is for everyone

Active projects

Unique repos with open help-wanted issues updated in the last 30 days

Showing 30 of ~22,702 matching filters (10+ stars)

List

Address race in PerformanceCounterLib

AI summary

A test in the `PerformanceCounterCategory_GetCounters` test suite is failing on an ARM64 Windows build. The error indicates that a required performance counter category, 'PerformanceCounterCategory_GetCounters_Category', does not exist, suggesting a potential race condition or setup issue during test execution.

arch-arm64os-windowshelp wantedblocking-clean-ci-optionalneeds-further-triagein-pr
18.0K
Difficulty
3/5

[reviewer] Review backlog (style): Destination-merge guard directory derived independently in boot and automerge; helper body duplicated inline

AI summary

This issue identifies a style inconsistency in the Hive codebase where the logic for deriving the destination-merge guard directory is duplicated in both the boot and automerge components. This duplication, along with an inlined helper function, creates a potential for silent failures if directory structures change. The proposed solution is to create a single exported helper function to be used by both components.

help wantedagent/reviewerblocked
51
Difficulty
2/5
glsec/glsec

GL005: GitLab's gl-secret-detection-report.json artifact flagged as a sensitive file

AI summary

GitLab's Secret Detection report artifact (`gl-secret-detection-report.json`) is incorrectly flagged as a sensitive file by the GL005 rule due to a broad `*secret*` pattern. This issue proposes excluding specific GitLab report artifacts from this pattern to eliminate false positives while ensuring genuine sensitive files continue to be detected.

buggood first issuerule
22
Difficulty
2/5

🐝 Hive Advisory Report

AI summary

This issue serves as a central hub for advisory findings from Hive agents, particularly at lower ACMM levels where agents analyze code without creating direct issues or PRs. A governor agent periodically posts digest comments summarizing these advisory findings. This issue is intended to be a living document and should not be closed.

help wantedagent/scannerhive/advisoryagent/security
119
Difficulty
2/5

[UI migration] Drop BlockedRoute

AI summary

This issue proposes the removal of the `BlockedRoute` component from the UI migration project. The component is no longer in use within the codebase, making its removal a straightforward cleanup task.

Good first issueHelp wantedDifficulty: EasyUI Migration
1.4K
Difficulty
1/5

Add support for bitbucket

AI summary

This issue requests the implementation of a new Git provider for Bitbucket. It specifically mentions building upon the new structural changes introduced in a recent pull request, indicating a need to integrate with an existing framework.

good first issueSize: Steam:library-maintainers
731
Difficulty
3/5

Add setting to prevent burrito from destroying resources

AI summary

This issue proposes adding a new configuration setting to the Burrito tool. This setting would prevent Burrito from executing 'apply' operations if the 'plan' output indicates that any resources are scheduled for destruction. The goal is to enhance safety and facilitate adoption on critical Terraform code by preventing accidental resource deletion.

good first issueSize: Steam:library-maintainers
731
Difficulty
2/5

Flags not saved

AI summary

Users are reporting that flags not connected to a road are not being saved and disappear upon loading a savegame. This prevents them from using standalone flags for in-game signaling, such as marking depleted mining areas.

help wanted
49
Difficulty
3/5

v0.53.1: Live kv_cache_type wiring on serve backends (vLLM / SGLang / transformers)

AI summary

This issue addresses a bug in v0.53.0 where the KV cache type setting for serving backends (vLLM, SGLang, transformers) is not correctly implemented, leading to a NotImplementedError. The proposed fix involves wiring the KV cache type through the CLI and into each backend, along with hardware capability checks for FP8 and a review of existing policies.

enhancementhelp wantedinfra-blockedclaimeddifficulty:7
72
Difficulty
3/5

VRAM pre-flight: price `backend: unsloth` from what Soup's Unsloth path builds (LoRA targets, embedding dtype)

AI summary

This issue addresses an inaccuracy in the VRAM estimation for the `backend: unsloth` configuration. The current estimation incorrectly uses default LoRA targets and assumes FP32 for embeddings/output heads, leading to inflated VRAM requirements. The proposed fix involves providing the backend to the estimator and dynamically determining LoRA targets and embedding dtypes specific to Unsloth.

enhancementhelp wantedclaimeddifficulty:7area:training
72
Difficulty
4/5

`soup eval gate`: an empty suite file prints pydantic's raw validation text without the file name

AI summary

When an empty YAML file is used with `soup eval gate`, the resulting Pydantic validation error displays raw Pydantic text instead of a user-friendly message that includes the filename and specific validation issues. The proposed fix involves catching Pydantic's `ValidationError` and re-raising a `ValueError` with a more informative message.

buggood first issueclaimeddifficulty:2area:eval
72
Difficulty
2/5

`soup train --resume`: "No checkpoint found to resume from." does not say which path or directory was searched

AI summary

The `soup train --resume` command provides a generic error message when no checkpoint is found, making it difficult for users to understand why. This issue proposes enhancing the error message to specify the exact path or directory that was searched and why it failed, improving user experience and debugging.

enhancementgood first issueclaimeddifficulty:3area:cli
72
Difficulty
2/5

Add native GRPO/RLVR training on the MLX backend

AI summary

This issue proposes adding native GRPO (Generative Reinforcement Learning with Proximal Optimization) and verifiable rewards training support to the MLX backend. Currently, MLX only supports SFT (Supervised Fine-Tuning) for these tasks, and GRPO training on MLX raises a NotImplementedError. The solution involves implementing a GRPO training loop for MLX, reusing existing reward functions and helpers.

enhancementhelp wantedclaimeddifficulty:10area:mlx
72
Difficulty
4/5

Test all major platforms

AI summary

This issue tracks the need to implement and run tests across various operating systems and architectures, including Windows (x86 and x64), Linux (x64 and ARM 64-bit), and macOS (M-series ARM). Currently, only Linux x64 has both manual and CI tests completed, with other platforms requiring further testing.

help wanted
26
Difficulty
3/5

To-do list for the new frontend

AI summary

This issue is a comprehensive to-do list for the development of a new frontend for the Movary application. It outlines the creation of various pages for movie management, user settings, and new features, with several items already marked as complete.

enhancementgood first issuepriority: high
773
Difficulty
4/5

Multi-Asset Candle Feed Not Working Reliably - get_candles_live()

AI summary

The user is experiencing unreliable candle data retrieval when requesting live, closed candles for multiple assets simultaneously using `get_candles_live()`. They suspect issues with rate limits, API limitations, connection handling, or data feed subscriptions, and are seeking confirmation on multi-asset support and best practices for their use case.

bugenhancementhelp wanted
111
Difficulty
3/5
genotrance/px

IPv6: getsockaddrarg() takes exactly 2 arguments (4 given)

AI summary

The issue describes a problem where a proxy (Px) fails to connect to IPv6 addresses, resulting in a 408 Request Timeout error. This occurs because the `getsockaddrarg()` function is being called with an incorrect number of arguments when resolving hostnames to IPv6 addresses.

enhancementhelp wanted
1.1K
Difficulty
3/5
genotrance/px

hostonly and WSL2

AI summary

The user is encountering issues using `px` with WSL2, specifically with the `--hostonly` flag not preventing connections from WSL2's internal network. They are seeing a "Client not allowed" error despite `proxy:hostonly` being set to 1, suggesting a potential problem with how `px` identifies and restricts access from WSL2's virtual network interfaces. The user suggests an enhancement to automatically include these virtual adapters as local.

enhancementhelp wanted
1.1K
Difficulty
3/5

Experiences with NovaCustom dGPU variants

AI summary

The user is seeking confirmation on whether NovaCustom NV56/NV54 laptops with dedicated GPUs boot and function correctly with Heads. They are aware of potential coreboot support and are willing to become a maintainer if the hardware is confirmed to be functional with Qubes OS.

help wantedBounty
1.6K
Difficulty
3/5

Password Reset via CLI does not work on Embedded MariaDB

AI summary

The password reset CLI tool in Uptime Kuma (version 2.0.0-beta.1) fails to prompt for a new password when using the embedded MariaDB database. This issue is specific to MariaDB, as the tool functions correctly with SQLite. The logs indicate an error related to locking the aria control file during MariaDB startup.

buggood first issueA:deployment
88.6K
Difficulty
3/5

Support multiple UI languages

AI summary

This issue proposes to internationalize the application's user interface by moving all English UI text from hardcoded strings into .resx resource files. The goal is to support multiple languages, with automatic language selection based on system settings and English as a fallback. The initial focus is on German translation, with subsequent languages to be added via machine translation and community review.

enhancementhelp wantedquestion
59
Difficulty
3/5

Review the Top 10 against NIST SP 800-239 (AI Data Center Security Analysis)

AI summary

This issue requests the creation of a crosswalk mapping the "Top 10" FORGE risks to relevant controls and subcategories within NIST SP 800-53, NIST Cybersecurity Framework (CSF 2.0), and the NIST AI Risk Management Framework (AI RMF). The goal is to help organizations align the FORGE framework with their existing NIST compliance programs by providing clear, defensible mappings and noting any gaps.

enhancementhelp wanted
19
Difficulty
3/5

Remove usage of AsyncResponse from SamlService

AI summary

This issue proposes removing the usage of AsyncResponse from the SamlService in Keycloak. The current implementation is complex, brittle, and can lead to issues with JTA transactions and thread-local management, potentially causing memory leaks. The suggested solutions involve either reverting to synchronous logic or introducing a new task runner that returns CompletableFuture<Response>.

area/samlkind/bughelp wantedstatus/auto-bumppriority/normalteam/core-protocols
35.5K
Difficulty
4/5
Expensify/App

[$250] Reports - Back button on report view returns to different scroll position

AI summary

When navigating back from an expense report on staging.new.expensify.com, the scroll position is not maintained, and the user is returned to a different, earlier scroll position. This issue occurs when an account has more than 50 reports and requires users to manually re-scroll to find their previously viewed report.

ExternalEngineeringDailyHelp WantedOverdueBug
4.9K
Difficulty
3/5

wp-admin autologin from OpenAdmin to site with moved wp-config.php is not working

AI summary

The autologin feature from OpenPanel's admin panel to a WordPress site with a moved wp-config.php file is failing. The error message indicates that the wp-config.php file cannot be found in the expected location, even though autologin works from the user panel. This suggests a configuration or path resolution issue specific to the admin panel's interaction with hardened WordPress installations.

bugenhancementgood first issue
688
Difficulty
3/5

[AIP-94] airflowctl tasks: add TasksOperations and TaskInstancesOperations to operations.py

AI summary

This issue proposes adding two new classes, TasksOperations and TaskInstancesOperations, to the airflowctl API. These classes will directly expose REST API responses without transformation, enabling auto-generation of CLI commands and unblocking future custom task command features. All necessary REST API endpoints are confirmed to exist.

area:CLIkind:featurearea:APIgood first issuekind:metaarea:airflow-ctl
46.0K
Difficulty
2/5

[good first issue, hacktoberfest] 🪭 Hacktoberfest Task - Add new Video Game Quote 641 (good-first-issue)

AI summary

This issue is a beginner-friendly task for Hacktoberfest, requiring contributors to add a new video game quote from Steins;Gate to a JSON file. No coding knowledge or local setup is needed, as the contribution can be made directly through the GitHub web interface.

help wantedgood first issuehacktoberfestcommunitylow hanging fruitup-for-grabs
1.9K
Difficulty
1/5
JabRef/jabref

Refine "Search for unlinked files" dialog

AI summary

This issue requests several user experience improvements for the "Search for unlinked files" dialog in JabRef. Key changes include styling the dialog to match JabRef's CSS, removing an unnecessary question mark icon, renaming a dropdown option, and improving keyboard navigation for the file tree to select/deselect checkboxes with the spacebar.

component: external-filesgood first issue📌 Pinned
4.4K
Difficulty
2/5

Enhancement: Query result pagination with cached result sets

AI summary

This enhancement proposes adding pagination to query results in the 'lore' CLI. Instead of re-running the entire query pipeline for subsequent pages, the CLI would cache the full result set and serve subsequent pages directly from the cache, significantly speeding up browsing large result sets. Cache invalidation would be tied to Git HEAD changes.

enhancementhelp wanteddraft
32
Difficulty
3/5
JabRef/jabref

Refine "Add JabRef suggested groups"

AI summary

This issue proposes refining the "Add JabRef suggested groups" feature to display a hierarchical tree structure of suggested groups, similar to the provided example. It requires understanding JabRef's group types, particularly free-form search expressions, to implement this visual enhancement.

component: groupsgood first issue
4.4K
Difficulty
3/5

How it works

Three steps from browsing to opening a PR.

Read the beginner guide →
  1. 1

    Filter the feed

    Pick a language, minimum stars, category, and difficulty that match what you can ship.

  2. 2

    Read the AI summary

    Each card includes a short summary and difficulty score when we have one — so you can skip the wall of text.

  3. 3

    Open the issue

    Jump to GitHub, talk to the maintainer if needed, and contribute where help is actually wanted.

Not finding what you’re looking for?

Browse 119,969 projects or reset filters to see more of the 833,825 indexed issues.