Loading...
Loading...
Found 1,285 Skills
Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.
Generate tone-of-voice and brand-system files for use with the PPTX Generator and other brand-aware skills. TRIGGERS - Use this skill when user says: - "help me create a brand system" / "set up my brand" - "generate my tone of voice" / "create my voice guidelines" - "configure my brand for presentations" / "set up brand for slides" - "create brand files" / "initialize my brand" - Any request about defining brand identity, voice, or visual system for content creation Creates brand.json, config.json, brand-system.md, and tone-of-voice.md files.
Add "Sign in with Apple" to an Ionic Capacitor app via @capacitor-community/apple-sign-in. Trigger when adding social login on iOS, integrating Apple ID as a sign-in method, or when App Review requires Sign in with Apple alongside other social logins.
PRD Expert Advisor Mode. Three well-known PM/CEO act as advisors, proactively review PRD, identify weak points and provide improvement solutions that can be directly written into the document. Users decide whether to adopt via accept/reject. Trigger methods: /li-prd-review, "Enhance PRD", "Expert Advisor", "Help me review this PRD", "Let experts review the PRD", "What are the loopholes in the PRD", "Help me optimize the PRD", "What other issues are there with the PRD". Even if users don't mention "PRD", this skill should be triggered as long as they want expert roles to review a product document and provide improvement solutions. DO NOT trigger for: PRD generation tasks (use li-prd), content script generation, topic analysis and other content creation tasks.
Smart Parking Open Platform · Parking Lot Domain (park): Query parking lot/parking lot list, basic parking lot information, parking lot system information, parking lot areas, channel information, cloud parking lot equipment, empty parking spaces/remaining parking spaces, empty parking spaces within an area, remaining parking spaces and free parking duration, parking lot fee information (fee inquiry), fee calculation for other vehicle types, free parking information for vehicles, vehicle display and voice prompts, vehicle coupon/e-coupon records, authorized parking lot codes, set real-time parking spaces. Trigger words: query parking lot, parking lot information, parking lot list, parking lot code, parkCode, empty parking space, remaining parking space, available parking space, free duration, fee inquiry, charging standard, fee calculation, vehicle type charging, free parking, display and voice, coupon, e-coupon, authorized parking lot, channel information, cloud parking lot equipment, area information, set parking space.
Primarily the agent's internal-thinking skill — invoke it silently to model a problem, identify trade-offs, and decide what to do, BEFORE asking the user anything or dispatching another skill. Workflow skills call `/culture` as their step-1 reasoning pass; the agent does not surface the dialogue. Only treat this as a user-facing skill when the user has explicitly opted out of writes — phrases like "no writes", "just rubber-duck this", "let's only talk", "/culture". In the user-facing path the output is conversation; the only sanctioned artifact is an opt-in `.cheese/notes/<slug>.md` handoff slug at session end if the user asks for notes. Culture never writes to production code, never commits, never opens PRs. If the dialogue reveals real work, recommend `/mold` (fuzzy → spec) or `/cook` (clear ask → code) and stop. Before `/mold` or `/cook`.
Search Mobbin for real app UI screenshots and visually analyze them. Required before calling the `search_screens` MCP tool — this skill defines how to plan searches, respond, and build HTML evidence boards when the screens are the answer. Use whenever the user asks about UI/UX design patterns, wants to see how other apps handle a screen or flow, needs design inspiration or references, asks to compare UI approaches across apps, mentions Mobbin, or whenever `search_screens` would be relevant. Trigger aggressively for any design-related question — even if screenshots aren't explicitly requested.
Hand off the current task to the SLICC browser agent, or install a new skill into SLICC from a GitHub repo. Use this skill when the user says things like "handoff to slicc", "move this to slicc", "move to the browser", "test in the browser", "handoff to browser", "install this skill in slicc", "upskill slicc with this repo", "add this skill to slicc", or otherwise asks you to continue the work inside the SLICC browser agent.
Owns the smoke test contract for an ML experiment: a small, diagnostic-by-construction pytest that fits the experiment's learner on a portion of the real `data/` source and predicts on a *disjoint* portion that deliberately carries **no pre-history buffer**. The assertion is structural — the number of predictions must equal the number of rows in the predict grid. A pipeline that loads-then-features-then-splits will silently drop the cold-start rows of the predict slice and the test will fail with a row-count mismatch; a pipeline that marks X early and references upstream history nodes from feature steps will pass trivially. The smoke test is the executable proof of the X-marker placement rule from `build-ml-pipeline`. TRIGGER when: `test-ml-pipeline` has dispatched here to write the smoke test for an approved experiment; `pytest tests/smoke/` is failing on row count; the user asks "why is the smoke test failing?"; a pipeline edit in `build-ml-pipeline` needs an executable proof; an experiment script changes the pipeline shape and the matching smoke test needs revisiting. SKIP when: the design note does not exist or is not yet approved (route to `iterate-ml-experiment`); the user is asking about a regression test or schema invariant (route to `regression-test-ml-pipeline` / `distribution-test-ml-pipeline` once those exist); the question is the *interpretation* of CV metrics, not predict-time correctness (route to `evaluate-ml-pipeline`). HOW TO USE: read the matching experiment's `journal/NN_*.md` and `experiments/NN_*.py` first to understand the pipeline's source binding (what env-dict keys does `build_learner` expect?). Then construct two env-dicts from the **real `data/` source** — a train env and a predict env — such that the predict env carries *only the rows we want predictions for* and *no pre-history buffer*. The hard assertion is that the prediction count matches the predict-env row count exactly. The soft assertion is that the smoke set's MAE is within `3 × CV_mean` (or the task-appropriate analogue). **Do not write the design note or run CV — that's other skills' job.**
Use when downloading videos, audio, or captions from YouTube and other video platforms. Supports quality selection.
Add server-side route protection to enforce authentication on specific pages while keeping others public.
Orchestrates complete project initialization by coordinating agent-folder-init, linter-formatter-init, husky-test-coverage, and other setup skills. Use this skill when starting a new project that needs full AI-first development infrastructure with code quality enforcement.