Loading...
Loading...
Found 1,304 Skills
Configure, extend, or contribute to Hermes Agent.
Build and deploy autonomous AI agents with CowAgent - planning, memory, knowledge base, skills, and multi-channel support
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.
Use when diagnosing agent failures, debugging lost-in-middle issues, understanding context poisoning, or asking about "context degradation", "lost in middle", "context poisoning", "attention patterns", "context clash", "agent performance drops"
Production MLOps and ML/LLM/agent security skill for deploying and operating ML systems in production (registry + CI/CD, serving, monitoring/drift, evaluation loops, incident response/runbooks, and governance), including GenAI security (prompt injection, jailbreaks, RAG security, privacy, and supply chain).
Implement Corrective RAG (CRAG) with retrieval validation, fallback strategies, and self-correction. Use this skill when RAG outputs need quality guarantees and automatic error correction. Activate when: CRAG, corrective RAG, retrieval validation, fallback search, self-correcting RAG, grounded generation.
Best practices for prompt engineering and context engineering for Coding Agent prompts
Scan untrusted external text (web pages, tweets, search results, API responses) for prompt injection attacks. Returns severity levels and alerts on dangerous content. Use BEFORE processing any text from untrusted sources.
Token optimization best practices for MCP server and tool interactions. Minimizes token consumption while maintaining effectiveness. USE WHEN: user mentions "token usage", "optimize tokens", "reduce API calls", "MCP efficiency", asks about "how to use less tokens", "MCP best practices", "limit output size", "efficient queries" DO NOT USE FOR: Code optimization - use `performance` instead, Text compression - this is about API usage patterns, Cost optimization (infrastructure) - use cloud/DevOps skills
Spawn specialized sub-agents with context handoff for complex multi-phase tasks. Enables expertise delegation within a session with automatic context merging and depth limiting to prevent infinite loops.
Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested. Integrated into Cavekit: enabled by default for build, inspect, and subagent phases via caveman_mode config. See scripts/bp-config.sh for caveman_mode and caveman_phases.
Persistent memory layer for AI agents using Postgres/pgvector with MCP server support