Recently updated open-source projects

spacex-daily-briefing turns live X and web research into a compact daily brief on SpaceX, its SPCX public entity, and the merged xAI business. It reads the previous brief, reports only the delta, and organizes price-moving signals, key metrics, segment watch items, catalysts, risks, and inline sources. The skill separates confirmed facts from chatter and refuses to invent figures or quotes. It is informational only and not financial advice.

Obsidian Project Documentation Manager captures project progress and decisions while you work with Claude Code, then updates a structured vault using customizable templates. It infers the project from the working directory, organizes related projects and technologies, and can optionally commit or push vault changes to GitHub for backup.

Elicitation Widget gives Claude skills a shared intake engine instead of making each one ask questions in prose. A skill declares a QUESTIONS array with unique keys, complete questions, and two to four options; the widget presents them in one native choice-card interaction and returns a structured key-value payload. The entire implementation lives in one SKILL.md, but it requires an environment that supports ask_user_input_v0 and does not silently fall back when that UI is unavailable.

relief-plate converts a high-resolution black-and-white PNG into an STL for relief printing: dark regions become raised ink surfaces and white regions remain recessed. Oversized artwork can be divided into tiles with hidden bottom dovetails so the printed relief joins across the seam, while detail test tiles expose fragile areas before a full print. The rules target a 0.4 mm nozzle and a Bambu X2D bed, and the scripts can run without the agent skill.

icon-generator creates project marks, app icons, and favicons by starting with bold, editable SVG concepts and escalating to image rendering only for dimensional or illustrative work. It presents two or three concepts in a contact sheet that includes tiny previews, then slices the chosen master into SVG, ICO, PNG, manifest, and HTML-head files. Standalone scripts check contrast and build the set. The full conversion path targets macOS tools such as sips and qlmanage, while the SVG-first route needs no rendering account.

mutation-killing-orchestrator reads cargo-mutants survivors, ranks them by silent-regression severity, domain criticality, blast radius, cluster size, and whether a test can distinguish the mutant, then selects up to three findings. A stronger model performs the judgment and cheaper agents write one killing test per finding before a scoped re-run verifies the result. Equivalent or low-value survivors are accepted or deferred instead of chasing a perfect score, and the workflow leaves tests staged rather than committed.

shipmate helps maintainers release JavaScript, TypeScript, or Python repositories locally, focusing on why a change matters rather than copying commit subjects into a changelog. It discovers version locations, handles repositories with multiple independent contracts, checks for drift, and pauses at a human checkpoint before tagging and creating the GitHub release. It is aimed at deliberate manual releases, not fully automated commit-driven CI.

durable-product-playbook helps a person or small team preserve product boundaries, done criteria, execution habits, and design-system decisions while working with an AI coding partner over many sessions. It includes reusable templates for a project manual, definition of done with drift guards, and a deployment runbook. The skill is for durable product practice, not a one-off deploy command, file summary, or throwaway prototype.

lark-todo is an agent skill that scans messages, meetings, calendars, document comments, approvals, tasks, and mail across one or more Lark organizations, then deduplicates and ranks the pending work by urgency. After the user selects an item, it can draft replies, process approvals, answer mail, respond to invitations, or create tasks, with every write action shown for confirmation first. Initial setup requires Node.js, a Lark custom app, and authorization through lark-cli.

servo structures agent development around named sources of truth and reviewers that did not produce the work. Its skills scan the project, gather relevant context, convene experts and a dissenter, shape a spec, seal a deterministic plan, watch scope during execution, close acceptance criteria, and run a non-colluding gate that returns GO, FIX, or STOP. A manifest records where project truth lives and who owns each check. The plugin is alpha; heavier backtesting and differential plan tools remain on the roadmap.

learn-from-project reads an AI-built or unfamiliar repository and writes a LEARNING.md that explains what it does, maps its structure, and walks through core files in execution order. It separates logic that can change a result from environment or presentation details, explains the stack for non-engineers, and compares claims with actual behavior. The guide ends with an explain-back and a short list of gaps to close; output is English unless another language is requested.

Citation Verification makes an AI assistant check a DOI, PMID, or arXiv identifier, read the abstract, and judge whether the paper supports the intended claim before citing it. Full text is fetched only when specific numbers or methods require it. The output includes bibliographic details, a conclusion, and a short quotation, or marks an inaccessible abstract instead of inventing one.

Spoiler-Free Replay finds an official full-match football replay on Xiaohongshu, opens it from kickoff in fullscreen, and blocks score-bearing highlights, titles, comments, and other visible spoilers. The workflow first confirms that the match has ended without telling the user the result, then checks the player view before handing it over. It depends on Claude in Chrome and a logged-in Xiaohongshu web session, and its controls are currently platform-specific.

Use this Claude Code skill to turn scripts into single-voice narration or multi-speaker podcast audio through ElevenLabs, including character-level timing for karaoke-style subtitles. Its cookbook focuses on practical Chinese pronunciation, pacing, emotion tags, chunking, MP3 merging, and browser synchronization issues. An ElevenLabs account and API key are required, and ffmpeg is needed for dialogue assembly and multi-chunk merging.

Use this Claude Code skill when selecting a chart, reviewing dashboard code, checking whether a graphic is misleading, or planning a map. It guides the response through data analysis, visual encoding, integrity checks, and cleanup, drawing on Bertin and Tufte while loading deeper references for color, accessibility, statistical graphics, and modern chart types only when needed.

Give independent reviewers the artifact without your expected verdict, then cycle through critique, adjudication, and synthesis while logging rejected findings. The method distinguishes model and perspective independence, reserves one round for a blind alternative, and stops by a precommitted rule rather than a feeling. Agreement still does not prove code, empirical, financial, or security claims.

journal-this turns a current work session into a dated Markdown journal entry written in the user’s voice. It reads recent entries for continuity and focuses on what changed or moved forward instead of producing a generic recap. Entries can stay in a local folder or be committed and pushed to a GitHub repository, with setup options and pull-rebase handling for concurrent updates.

Give the skill a life-hack direction and it develops the concept, creates a storyboard, generates images and video clips, and assembles a 9:16 final cut with AI voiceover. It uses Volcengine ARK models for generation and local ffmpeg for stitching, while also delivering source clips, storyboard assets, and a voiceover script. Dialogue-heavy or performance-led videos are outside its main scope.