Create, publish and manage classic LinkedIn job postings with the Curviate CLI, and read their applicants. Covers `job` (get, list, create, update, budget, publish, close) and `job applicants` / `job applicant get|resume`, the draft-then-publish flow, the subscription gate that stops a publish even on the free mode, and the read-back delay after an update. Use when drafting or updating a posting, pricing or publishing one, closing one, or pulling applicants and résumés.
curviate1 file · MIT
Grow and manage a LinkedIn network with the Curviate CLI. Covers `connect` (send with a note, sent, received, accept, decline, cancel), `profile follow`/`unfollow`, `profile relations`, `profile followers` and `profile following`, the de-duplication signal for a connect loop, and the invitation propagation delay. Use when sending or withdrawing connection requests, triaging received invitations, following a member, or listing connections and followers.
curviate1 file · MIT
Drive LinkedIn Sales Navigator and Recruiter through the Curviate CLI. Covers `sales-nav` (search, profile, message, lead and account lists, save-lead, save-account) and `recruiter` (projects, pipeline, talent-search, save-candidate, project jobs, applicants, message, profile, search), and what can refuse them: the LinkedIn subscription on the connected account, the Curviate seat, and the beta-operations gate. These commands carry no Curviate add-on paywall. Use when working a Sales Navigator or
curviate1 file · MIT
Read and write LinkedIn profiles and company pages with the Curviate CLI, and resolve human search terms into the opaque filter ids the search commands need. Covers `profile` (me, detail, sections, update, subscription, analytics, visitors, SSI), `company` (detail, employees, posts, jobs, followers, page inbox, follow-invite), `search parameters`, retrieval mode (`--mode`/`--max-age`), and the session/account commands (`login`, `config`, `account`). Use when fetching or updating a member or comp
curviate1 file · MIT
The first run: install the Curviate CLI, authenticate this machine with `curviate setup`, confirm with `curviate doctor`, connect a LinkedIn account, and prove the whole path works by reading the same profile live and then from the store. Use when Curviate is being set up for the first time, when `curviate` is not installed or not authenticated, when a command fails with a credential or account error, or when someone asks how to get started. Reads only: it issues no write of any kind.
curviate1 file · MIT
Find people, companies, posts, jobs, service providers and groups on LinkedIn with the Curviate CLI. Covers `search people|companies|posts|jobs|services|groups`, running a pasted LinkedIn search URL directly, the `group` read commands, pagination with `--all`, and the filter traps that silently return unfiltered results. Use when sourcing prospects or candidates, qualifying companies, finding posts or job postings to engage with, or resolving a group.
curviate1 file · MIT
GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora.
ai research3 files · MIT
Scalable data processing for ML workloads. Streaming execution across CPU/GPU, supports Parquet/CSV/JSON/images. Integrates with Ray Train, PyTorch, TensorFlow. Scales from single machine to 100s of nodes. Use for batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines.
ai research3 files · MIT
Add official Railway database services (Postgres, Redis, MySQL, MongoDB). Use when user wants to add a database, says "add postgres", "add redis", "add database", "connect to database", or "wire up the database". For other templates (Ghost, Strapi, n8n), use the railway-templates skill.
railway1 file · MIT
Design robust, scalable database schemas for SQL and NoSQL databases. Provides normalization guidelines, indexing strategies, migration patterns, constraint design, and performance optimization. Ensures data integrity, query performance, and maintainable data models.
development4 files · MIT
Review and manage Dependabot PRs. Categorizes by risk, checks CI status, auto-merges safe updates, and reports issues. Use when the user says "review dependabot", "merge dependabot", "dependabot PRs", or "update dependencies".
workflow automation1 file · MIT
Smart dependency management for any language. Auto-detects project type, applies safe updates automatically, prompts for major versions, diagnoses and fixes dependency issues.
development4 files · MIT
Deploy code to Railway using "railway up". Use when user wants to push code, says "railway up", "deploy", "ship", or "push". For initial setup or creating services, use railway-new skill. For Docker images, use railway-environment skill.
railway1 file · MIT
Manage Railway deployments - view logs, redeploy, restart, or remove deployments. Use for deployment lifecycle (remove, stop, redeploy, restart), deployment visibility (list, status, history), and troubleshooting (logs, errors, failures, crashes). NOT for deleting services - use railway-environment skill with isDeleted for that.
railway1 file · MIT
Create and evolve design systems with design tokens, component architecture, accessibility guidelines, and documentation templates. Ensures consistent, scalable, and accessible UI across products.
development6 files · MIT
Use when Codex is building or iterating on a web game (HTML/JS) and needs a reliable development + testing loop: implement small changes, run a Playwright-based test script with short input bursts and intentional pauses, inspect screenshots/text, and review console errors with render_game_to_text.
creative design7 files · Apache-2.0
Reviews a product document (PRD, spec, design brief) BEFORE implementation to surface holes — undefined edge cases, missing states, policy gaps — by attacking what the document is SILENT about (things unwritten, and things written only for the happy path). Acts as a strict "sign-off manager," ruling Approve / Conditional / Reject and producing a polite, forwardable question list. Works for planners/PMs (self-review before sharing), engineers (blocking questions before coding), and designers (scr
productivity3 files · MIT
After you give a substantive answer or draft that the user may act on — advice or recommendations, drafted artifacts such as goals, plans, pitches, proposals, or emails, estimates or projections, analysis or interpretation of data, factual claims they may rely on, or a multi-step argument — invoke this skill BEFORE finalizing your reply and then, if it applies, append 2-3 short follow-up questions, each tied to something specific in what you just produced, that help the user check key facts, pro
productivity2 files · Apache-2.0
Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.
ai research4 files · MIT
Expert guidance for distributed training with DeepSpeed - ZeRO optimization stages, pipeline parallelism, FP16/BF16/FP8, 1-bit Adam, sparse attention
ai research10 files · MIT
Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies. Use when training models >1B parameters, need maximum GPU efficiency (47% MFU on H100), or require tensor/pipeline/sequence/context/expert parallelism. Production-ready framework used for Nemotron, LLaMA, DeepSeek.
ai research5 files · MIT
Expert guidance for Fully Sharded Data Parallel training with PyTorch FSDP - parameter sharding, mixed precision, CPU offloading, FSDP2
ai research3 files · MIT
High-level PyTorch framework with Trainer class, automatic distributed training (DDP/FSDP/DeepSpeed), callbacks system, and minimal boilerplate. Scales from laptop to supercomputer with same code. Use when you want clean training loops with built-in best practices.
ai research4 files · MIT
Distributed training orchestration across clusters. Scales PyTorch/TensorFlow/HuggingFace from laptop to 1000s of nodes. Built-in hyperparameter tuning with Ray Tune, fault tolerance, elastic scaling. Use when training massive models across multiple machines or running distributed hyperparameter sweeps.
ai research2 files · MIT
Use when the task involves reading, creating, or editing `.docx` documents, especially when formatting or layout fidelity matters; prefer `python-docx` plus the bundled `scripts/render_docx.py` for visual checks.
document processing6 files · Apache-2.0
Free lite skill — drafts one friendly (level-1) client document-request email for a single tax client from a short brief plus the practice profile. A strict subset of client-doc-chaser (no batch, no escalation levels 2/3), carrying the pack's full guard. Use to try the kit before buying, or as the directory-published funnel skill. Triggers - "client document request email", "chase a client's documents", "free document request writer", "doc-chaser-lite".
productivity9 files · MIT
Add, view, or remove domains for Railway services. Use when user wants to add a domain, generate a railway domain, check current domains, get the URL for a service, or remove a domain.
railway1 file · MIT
Compress large language models using knowledge distillation from teacher to student models. Use when deploying smaller models with retained performance, transferring GPT-4 capabilities to open-source models, or reducing inference costs. Covers temperature scaling, soft targets, reverse KLD, logit distillation, and MiniLLM training strategies.
ai research2 files · MIT
Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques. Use when processing long documents (32k-128k+ tokens), extending pre-trained models beyond original context limits, or implementing efficient positional encodings. Covers rotary embeddings, attention biases, interpolation methods, and extrapolation strategies for LLMs.
ai research4 files · MIT
Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.
ai research4 files · MIT
