Productivity
token-conservation
Enforces token quota management at session start with conservation and compression checks. Use at the start of every session or before large context loads
Productivity
response-compression
Compresses verbose responses by removing filler and framing to save 200-400 tokens
Productivity
context-optimization
Optimizes context window via MECW principles and memory tiering
AI / ML
openrouter-integration
Connect apps to 300+ AI models through OpenRouter's OpenAI-compatible API — model discovery, image generation, multimodal chat (text, images, PDFs), exact per-generation cost track…
AI / ML
factorized-video-gen
Factorized video generation pipeline based on arxiv:2512.16371. Decomposes text-to-video into three stages: (1) LLM reasoning to rewrite prompts into first-frame captions, (2) T2I …
AI / ML
artificial-analysis
Pick the right LLM or media model for a task, backed by live benchmark data from artificialanalysis.ai. Use when the user asks "which model should I use for X", "what's the best/fa…
Research
study_project
对开源大模型项目进行系统化分析：提取并翻译所有提示词（prompt），梳理工具定义和 LLM 调用架构，生成技术分析报告和面向非技术人员的产品说明书。当用户想要学习/分析/研究/调研一个 LLM/AI 项目、提取 prompt、查看项目用了什么提示词、理解大模型调用逻辑、做竞品技术调研、或为 AI 项目生成报告和产品说明书时，都应该使用这个 skill。即使…
Security
Alepha188838884/context-firewall
Local proxy that collapses N downstream MCP servers into 4 meta-tools with progressive tool discovery (measured: 122 tools → 4, ~28.6K tokens of definitions saved), compresses larg…
AI / ML
ClawRouter-Hermes
Adds [ClawRouter](https://github.com/BlockRunAI/ClawRouter) as a native model provider: 55+ LLMs from 9 providers through one local proxy, paid per-call in USDC via x402 instead of…
AI / ML
connect-llm
Connect an LLM or MCP client to Emisar and certify the connection end to end. Use for installing the emisar-mcp stdio bridge, registering a local client (Claude Code, Cursor, Zed, …
AI / ML
mistral-mcp
MCP server (stdio + Streamable HTTP) wrapping the full Mistral AI surface — chat, embeddings, vision, OCR, Voxtral audio (transcribe/speak), Codestral FIM, agents, moderation, clas…
DevOps
llm-gateway-incident-triage
Diagnostic playbook for triaging a self-hosted LLM gateway (LiteLLM or any Postgres-backed OpenAI-compatible proxy) running via Docker Compose on a homelab host. Use when a client …
AI / ML
prompt-engineer
Experto en diseño y optimización de prompts para LLMs. Domina estructura de prompts, gestión de contexto, formato de outputs, few-shot, chain-of-thought y evaluación sistemática. U…
Productivity
total-recall
The only memory skill that watches on its own. No database. No vectors. No manual saves. Just an LLM observer that compresses your conversations into prioritised notes, consolidate…
AI / ML
fabric-ai-functions
Fabric AI Functions skill enabling one-line LLM operations on pandas and PySpark DataFrames: sentiment, classification, extraction, embeddings, summarization, translation, grammar …
AI / ML
agento11y-instrument
Sets up and instruments LLM apps or agents to send generations and workflows to Grafana Agent Observability. Detects language/framework and instrumentation state, then runs gap che…
AI / ML
local-llm-router
Route AI coding queries to local LLMs in air-gapped networks. Integrates Serena MCP for semantic code understanding. Use when working offline, with local models (Ollama, LM Studio,…
AI / ML
forge-evals
Design evaluations for LLM features including golden datasets, rubric scoring, LLM-as-judge calibration, CI regression detection, online A/B tests, cost and latency budgets, and ad…
Research
speech-paper-daily
语音领域每日论文速递。搜索最新语音大模型(Speech LLM、TTS、ASR、codec、speech generation)和语音前端(speech enhancement、noise suppression、beamforming、source separation、dereverberation)预印本论文,以毒舌但判断极准的 senior revi…
AI / ML
promptl
PromptL syntax guide for writing prompts in the Latitude platform. This skill should be used when writing, reviewing, or editing PromptL prompts. Triggers on tasks involving creati…
Testing
agento11y-test-starter
Builds an offline test suite for an agent before deployment by analyzing its code, prompts, and tools to generate labeled cases (happy/edge/adversarial) with scoring recommendation…
Web
forge-llm-streaming
Streaming LLM output correctly with SSE setup, partial-token UTF-8 decoding, AbortController cancellation, mid-stream errors as events, append-only UI rendering, and backpressure h…
AI / ML
mllm-eval
Designs or audits a model-agnostic evaluation harness for LLMs or multimodal models on clinical tasks, covering reference standards, clinical metrics, faithfulness checks, contamin…
AI / ML
caucus
Multi-LLM consensus engine. Queries multiple AI models and synthesizes a single answer using strategies like voting, judge synthesis, and multi-round debate. Use when the user want…
AI / ML
agento11y
Inspects and manages Grafana Agent observability resources—conversations, generations, evaluators, rules, scores, and templates—via gcx. Use for listing or searching conversations,…
AI / ML
code-and-llm-collaboration
Designs hybrid pipelines blending deterministic code with LLM calls. Catalogs four proven layering patterns—LLM→Code guard, Code filter→LLM, LLM judge + Code enforce, Code orchestr…
AI / ML
PolyBrain
Multi-agent, multi-model orchestration skill that decomposes an objective into a JSON task plan, runs research and build subtasks in parallel, then synthesizes and verifies a final…
AI / ML
ship-helpbot
Builds an embeddable "ask the guide" assistant: chunks user-guide markdown into docs/guide-index.json, emits a framework-agnostic <ship-helpbot> web component, and a reference /api…
Productivity
prompt-optimizer
Use this skill when the user asks to optimize a prompt, generate the best prompt for a task, improve an existing prompt, or write a prompt from scratch. Triggers include: "optimize…
AI / ML
efficient-dispatch
Orchestrates model routing for frontier LLMs: keeps primary judgment on the main model while dispatching mechanical work to cheaper sub-models, then measures routing effectiveness.…
AI / ML
output-dev-model-selection
Pick the right LLM model for an Output SDK prompt file. Use when writing a new .prompt file, reviewing a model choice, or upgrading a stale model. Walks through priority (reasoning…
AI / ML
pixeltable
Declarative tables for multimodal AI apps: unifies chunking, embeddings, retrieval, and 25+ LLM providers through computed columns that trigger on insert. Handles RAG pipelines, me…
Research
Awesome llm agents
· ([mp.weixin.qq](https://mp.weixin.qq.com/s?__biz=MzAxMTk4NDkwNw==&mid=2247495643&idx=2&sn=f759dc738477d76143ffc53a043b398c&poc_token=HPiN6mWjzo5Wn8_C9o639UGw2ABkm8pA7X9RusYq)) - …
AI / ML
Clauddib Website
Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse tasks, yet their reasoning remains fundamentally linear—generating thoughts token-by-token wit…
Research
Cursor Cloud Agents & Parallel Worktree Agents
New Cursor Interface · Cursor Skip to content Cursor Product↓ Agents Code Review Cloud ↗ Tab CLI Marketplace ↗ Enterprise Pricing Resources↓ Changelog Blog Docs Community Help ↗ Wo…
AI / ML
Foundation Plr
It is clear you have built a sophisticated research pipeline, but "context amnesia" in LLMs often stems from context fragmentation—where the instructions are present but spread acr…
Research
Generated Non Audited Paperwork
大型语言模型（LLM）的出现正深刻地重塑知识工作的边界，其中以代码辅助工具（如GitHub Copilot）的广泛普及最为引人注目，并已成为行业主导范式。本文旨在挑战一个普遍存在的假设，即此种主导地位是技术与任务最优对齐的必然结果。我们提出一个核心论点：LLM在编程领域的统治地位，是一个由精准的市场进入策略、随之产生的路径依赖以及深刻的社会学因素共同塑造的产…
Engineering
Rawrxda
RawrXD is a multi-target AI/LLM development environment with three main executables: - **RawrXD-QtShell**: Qt-based IDE with integrated GGUF model loading, inference, and autonomou…
AI / ML
hermes-model
Configure the Hermes model and provider with an API key over SSH, validate context window, and detect silent capability degradation. Use when setting up or changing which LLM a rem…
General
sales-serplock
Tracks LLM visibility via Prompt Rank, monitors brand perception in Brand Wiki, maps entity relationships in Topic Graph, and generates brand-aligned AI content. Handles Reddit mon…
AI / ML
agent-incident-postmortem
Conducts a blameless postmortem on AI agent or LLM failures—hallucinations, runaway tool use, prompt injection, cost spikes, or unintended autonomous actions—producing structured t…
AI / ML
annotate-spans
Records structured annotations on agent spans and traces, coaches review practices, and supports failure taxonomies or human/LLM evaluation workflows. Load when preparing to save f…
AI / ML
Multi Llm Wrapper
<task> Current-Sub-Project: Brave Search Knowledge Aggregator Mode: Code Directory: c:/dev/od-dm-y/OneDrive/dev-projects/multi-llm-wrapper/src/brave_search_aggregator Docs_Director…
AI / ML
scholar-annotate
Scales unstructured text into validated, structured variables via LLM annotation: codebook design, gold-set construction, prompt optimization, κ ≥ 0.70 reliability gate, and distil…
AI / ML
terminal--llamafile
Expert guidance for llamafile, the tool that packages LLMs into single executable files runnable on Linux, macOS, Windows, and FreeBSD without installation. Creates portable AI app…
Research
llm-citation-audit
Audits whether a brand is cited by LLMs for the prompts its buyers actually ask, identifies four failure modes (retrieval gap, entity gap, format gap, competitor moat), and deliver…
AI / ML
llm-assistant-hub
Optimizes long-context reasoning for high-risk documents in legal, commercial, and policy domains via layered analysis, chunked processing, assumption detection, structured compres…
AI / ML
model-profiler
Profiles models from any provider family against a fixed set of 10 work categories, mapping public benchmarks into tier rankings with full provenance. Emits routing-table.json, aud…
AI / ML
frag-cost
Analyzes finished run logs to measure whether session-key fragmentation triggers real cold-starts by comparing cache metrics on fresh-key versus repeat-key turns. Reports mean cach…
AI / ML
llm-deploy-training
Provides deployment and training for LLMs/VLMs: vLLM OpenAI-compatible endpoints, multimodal constraints, Qwen3.5 tool calling, reasoning controls, CUDA Graph tuning, plus ms-swift…
AI / ML
local-models
Run private, offline LLM tasks on local models via llama.cpp, reusing Ollama downloads. Supports summarization, classification, JSON extraction, PII anonymization, translation, emb…
AI / ML
medagentbench-setup
Deploys Stanford MedAgentBench for evaluating LLM agents in a simulated EHR/FHIR environment: clones the repo, sets up a Python 3.9 environment, starts HAPI FHIR with mock patient …
Automation
litcoin-standalone-miner
Autonomous LITCOIN ($LITCOIN) proof-of-research miner. Signs directly from your Base L2 wallet and uses OpenRouter for LLM inference to solve research tasks—agentic traces, instruc…
Web
make-wpf-chat-orchestrator
Builds a streaming chat orchestrator that converts IChatClient responses into UI-agnostic events (text, tool-started, tool-completed, failed) and integrates MCP tools via McpClient…
AI / ML
north-mini-code
Integrates Cohere North Mini Code 1.0 (30B MoE) into any coding agent: auto-detects hardware, picks optimal quantization, launches a local inference server, and routes requests wit…
AI / ML
hypogenic
Automated hypothesis generation and testing with LLMs. Generates scientific hypotheses from datasets, combines literature with empirical data, and validates them against observatio…
Web
make-wpf-chatclient-factory
Creates a reusable IChatClient factory sharing a single SocketsHttpHandler across OpenAI, Ollama, Anthropic, and Gemini providers, plus a MockChatClient for offline work. Ideal for…
Productivity
iceage
Ultra-compressed mode that slashes token use ~75% via terse, ice-age style speech while preserving technical precision. Supports intensity levels lite/full/ultra and wenyan variant…
AI / ML
llamafactory-sft
Guides a complete LlamaFactory SFT pipeline: data and model preparation, training configuration, background fine-tuning, loss visualization, validation, and export. Activates on re…
Productivity
llm-text-hygiene
Detects and removes LLM artifacts—prompt remnants, placeholders, AI thanks, and stylistic patterns—from finished documents (papers, READMEs, reports, blogs, applications). Validate…
Showing the top 60 of 1,648. See the full list →