Find the open-source solution that fits.

LoopCoding interviews the user and inspects the actual repository, installed skills, agents, connectors, hooks, workflows, and test commands before writing a LOOP.md. That specification defines the goal, machine-checkable done conditions, per-iteration checks, guardrails, escalation rules, and progress memory. A /loop runner or sub-agent then executes it until the checks pass. The approach is most useful when a task needs repeated changes and verification rather than prompt-by-prompt supervision.

Start with a topic, notes, finished material, or an existing PPTX, and the skill helps a coding agent shape the talk around one clear idea before building a fixed-stage HTML deck. It adds timed reveals, animation, optional sound, and inline editing, while scaling the 1920×1080 composition as one unit across venue screens. Scripts support deployment and PDF export; sound-enabled decks may use separate audio assets.

mule-headless-dev provides cross-platform scripts for building, deploying, hot-redeploying, restarting, and testing Mule 4 applications on a real standalone runtime without Anypoint Studio. The same scripts can be used as a Claude Code skill or directly from Windows, macOS, and Linux. Mule 4.9+ requires JDK 17 and Maven 3.9+, while enterprise runtime and connector downloads require the appropriate MuleSoft repository credentials in Maven settings.

paxel-skill evaluates local Claude Code and Codex CLI sessions against Paxel’s five axes: execution leverage, steering, engineering quality, product thinking, and planning. It reads project logs and Git history, removes tool bodies and secrets, extracts events and decisions, then asks Claude Haiku for narratives and scores without uploading data or using a backend. Results include a performance band, but the project warns that the model differs from YC’s scorer, reruns vary, and the overall score is only an approximation.

Product UI Design helps build dashboards, SaaS screens, settings, tables, forms, and admin tools with a restrained product register. It asks the agent to observe real products, separate functional restraint from expressive styling, and scan for common AI-looking choices before delivery. It is meant for product surfaces rather than marketing pages; an optional script checks machine-detectable tells.

Give Find Angles an unfocused topic and it produces a structured tree of distinct research directions, concrete sub-angles, keywords, and an academic search query for every node. The workflow stays anchored to the terms you supplied and can run through Claude, Codex, or another capable model without web access. It maps where to investigate; it does not find papers or claim that proposed findings already exist.

Owl Browser gives Claude Code a browser MCP server and an observe-and-act skill that renders pages as compact OwlMark text with stable handles. The agent navigates, clicks, and types by handle rather than guessing from screenshots or pixel coordinates. A running Owl Browser instance, endpoint and token environment variables, and npx are required before the plugin can operate.

ClickUp Operator organizes ideas, to-dos, and project work in ClickUp through a shared set of routing and setup rules. The package keeps its portable logic, templates, examples, and references in one core, then adds separate installation wrappers for OpenClaw or ClawHub and Claude Code. Maintainer notes and validation materials are kept apart from the runtime package, so the same operator behavior can be reused across both agent environments.

Product Modelling — Cloning an Influencer provides an eight-step method for turning a reference video into product-modeling stills and videos. It extracts frames, builds a character sheet, creates identity-anchored variants, adds products and storyboard shots, then makes a local fast-cut montage with frame-by-frame checks. The repository ships scripts and guidance only: no likeness, brand footage, or product photos are included. Users must supply licensed material and obtain the real person’s consent before cloning or publishing a likeness.

printable-docs helps create Traditional Chinese A4 reports, quotes, contracts, minutes, forms, resumes, slides, and study sheets from self-contained HTML. It includes 39 templates, a style contract, reusable grid CSS, offline SVG infographic modules, and a page-assembly tool. Its explicit embeddable font stack addresses the macOS Chrome problem where Chinese text can disappear in printed PDFs. Before delivery, the project recommends printing a real PDF and checking selectable text, page count, and table breaks.

macos-screen-recorder captures the main macOS display and the sound your Mac is playing, then writes a shareable H.264/AAC MP4 from a headless command-line run. It uses ScreenCaptureKit in one Swift source file, needs only the standard Screen Recording permission, and avoids virtual audio drivers or sudo. Duration is set when the command starts, and window or region selection is not included.

Prompt Framework Generator assembles prompts from a reusable skeleton, replaceable variables, and concrete style anchors. Its five-stage method chooses a structure, fills the variables, translates vague style words into actionable dimensions, emits Chinese and English versions, and checks the result. The same framework can support image generation, article writing, and video scripts instead of locking the structure to one content type.

Burnout Guard estimates burnout risk from self-report plus behavioral signals such as prompt heartbeats, late-night work, continuous blocks, and active-day streaks. Six levels range from quiet monitoring to single-task mode and a platform-level lockout with a cooldown, and Claude Code hooks can enforce the decision before a prompt reaches the model. State stays local, optional sleep and calendar inputs affect strain, and the project explicitly is not a medical device.

Use it to review essays, posts, documentation, or professional email before publication. The Claude Code skill identifies filler, stock structures, vague verbs, and over-compressed slogans, then rewrites the text with clearer subjects and more varied rhythm. It can be installed as a plugin or copied into a skills directory, but it does not check grammar or verify factual claims.

The skill gathers swim, bike, run, sleep, HRV, and related records from connected fitness MCP servers, normalizes and deduplicates them, then calculates training-load and recovery metrics with a local Python script. It returns a coach-style snapshot, discipline trends, safety flags, and a concrete next-session recommendation. At least one fitness data source is required, and the output is decision support rather than medical advice or a certified coach.

Run one command to turn a mixed track into instrumental, full-vocal, harmony, lead-vocal, extracted-reverb, dereverbed, and lightly de-essed outputs. The cross-platform Python driver uses several audio-separator model ensembles and can also be invoked as an agent skill. It requires Python, ffmpeg, and audio-separator, downloads roughly 5–6 GB of models on first use, and runs much more slowly without acceleration. The downloaded separation models have their own licenses, which should be reviewed separately for commercial use.

The Programmatic SEO Expert helps design scalable page systems with unique value for both Google Search and AI retrieval. It first tests whether programmatic SEO fits the business, then covers data foundations, server-rendered HTML, answer-first structure, index management, and organic plus AI visibility measurement. A staged rollout discourages launching huge page counts before quality is proven. References also cover GEO, AEO, LLMO, crawler access, and rendering checks, while fast-changing mechanics should be re-verified against current guidance.

Tribunal reviews only the current branch diff against a base branch. A hater examines each changed file, an integration reviewer looks for broken cross-module contracts, and a judge tests whether each accusation is deliberate or defensible before the verdict keeps the real weak spots. The skill writes a severity-ranked report to docs/reviews/ and works across common languages without external dependencies. It is not a general audit of the repository’s older, unchanged code.