Recently updated open-source projects

Aside occasionally identifies a cognitive bias or psychological pattern reflected in the user's own words and cites an academic source. It is designed as a quiet mid-conversation observation rather than a diagnosis, profile, or recommendation. The skill runs locally without telemetry or external data transfer, and a rejection or stop phrase disables it for the rest of the session.

Shakedown runs a whole-project review covering architecture, resilience, performance, storage waste, security, and documentation, then ranks findings by practical impact. It can be pointed at web apps, CLIs, agent systems, pipelines, microservices, or infrastructure. Teams may run the full Claude Code workflow or use individual reference checklists for a focused audit, making it complementary to tools centered on a single pull request.

When a user complains about a response, Bust Apologizer scores the miss from one to five and pairs a severity-shaped ASCII meter with a concise apology, an explanation, and a concrete fix. It is an adult-humor hobby plugin with a gate intended to suppress the joke in legal, professional, sensitive, or customer-facing situations.

Give pubmed-brief a biomedical topic and it searches PubMed, Europe PMC, and Crossref for recent and highly cited papers. It packages a cross-paper overview, structured summaries, citation details, and source links into a printable PDF; closed-access articles are summarized from abstracts when full text is unavailable.

Provide a finished DOCX, a reference-manager export, and one of eleven supported dental journals to receive a reformatted manuscript, a validation report, and an optional cover letter. The tool adjusts page setup, heading structure, citation rendering, and journal limits while preserving the manuscript text. It does not submit the paper, preserve tracked changes or comments, or insert the newly rendered bibliography back into the output DOCX in the current version.

Keep frontend changes aligned with Feature-Sliced Design by checking layer hierarchy, downward-only imports, slice-level public APIs, segment folders, and naming conventions. The skill can also scaffold a component or feature slice with the expected structure. It supports React Router, Next.js, Remix, and Vite React, with partial coverage for Astro's pages layer.

Use this Claude Code skill when starting, refactoring, or reviewing a full-stack application that needs consistent engineering defaults. Its workflow moves from schema and shared contracts to a layered repository, service, controller, and router implementation, while adding validation, centralized errors, migrations, indexes, security controls, frontend state patterns, and tests. The prescribed structure is intentionally heavier than needed for one-off scripts or disposable prototypes.

Prompt Injection Auditor reviews a system prompt for direct and indirect injection, role hijacking, instruction leakage, goal manipulation, memory poisoning, delimiter escapes, and multi-agent or tool-call attacks. It returns severity-ranked findings, concrete exploit examples, and a hardened replacement prompt without requiring external tools. Its scope is prompt-level security, so it does not detect application bugs, model-specific bypasses, steganographic attacks, probabilistic failures, or future zero-day jailbreaks.

Image Generation is a model-agnostic prompting guide for agents working with Imagen, Gemini Image, GPT Image, DALL·E, Stable Diffusion, Flux, Ideogram, Midjourney, and similar tools. It organizes prompts around subject, action, setting, composition, lighting, and style, with patterns for photography, illustration, logos, product shots, infographics, UI mockups, and image text. It also covers edit-preservation lists, failure diagnosis, and practical parameters such as aspect ratio, resolution, seed, and reference fidelity; it does not generate images by itself.

Bombaclaude changes Claude's conversational register to Jamaican Patois with lite, full, and explicitly opt-in raw tiers. The setting lasts until changed or the session ends, and a companion grill mode can stress-test plans in the same voice. Code blocks, paths, errors, command output, commits, and pull-request text stay untouched, while security warnings and destructive-action confirmations automatically return to plain language for clarity.

This agent skill connects Claude Code, Codex, and similar runtimes to the Fangxin API’s gpt-image-2 endpoint for text-to-image, image editing, and multi-image composition. It saves outputs inside the current project and can rotate backup keys on common upstream failures; a Fangxin API key is required, and high-resolution or multi-image jobs may take longer.

The toolkit incrementally downloads Feishu docx and wiki content to Markdown, uploads Markdown with real LaTeX equation blocks, and imports CSV, TSV, or Markdown tables into standalone Sheets. It works as a Claude Code skill, bash CLI, or agent tool, and requires a Feishu self-built app with the appropriate scopes and credentials.

Before any UI is produced, the skill requires a specific cultural register with real references rather than a vague style label, then seals the allowed colors, typography, and other tokens. Four gates check token use, prohibited patterns, hero hierarchy, and copy specificity, while ten starter registers and a companion task-first UX skill provide reusable starting points.

Claude Think Twice combines a reflection skill with a Claude Code pre-push hook. When a push is attempted, it inspects the pending file types, presents the corresponding minimum scan matrix, and asks the human whether to proceed, helping expose rushed assumptions before CI. The hook cannot verify that scans were actually run and is not a linter, test runner, or replacement for CI; its enforcement is the human confirmation moment.

Create agents, send tasks, inspect live activity, and collect structured results from a common terminal interface. SageCLI can switch or fail over among Claude Code, Codex, Gemini CLI, local models, and other supported runtimes while fitting into shell pipelines and CI. It is an orchestration layer rather than a coding assistant and relies on Bash, jq, and tmux.

emoji-speak converts explanatory prose in Claude Code into emoji-heavy responses while leaving code blocks, inline code, file paths, URLs, error messages, version numbers, and proper nouns unchanged. It can reduce output tokens for naturally verbose explanations, but the project's own benchmark notes that wrapping already-short answers in emoji may use more tokens instead.

Bulk PR Review maps dependency layers and cross-PR conflicts before dispatching parallel reviewers within each layer. It processes layers sequentially, gathers verdicts such as approve, request changes, blocked by CI, or close, and posts them directly through `gh pr review`. The workflow requires an authenticated GitHub CLI and `jq`.

This skill changes claude.ai's editing response into a ready-to-paste Python heredoc that replaces an exact old block with a new one, instead of printing the entire file again. A NOT FOUND guard prevents silent misses and can reduce output use on large files; it requires Python 3 in your terminal and is not intended for Claude Code's native file tools.