How this agent operates — its isolation, permissions, and tool access model
Agent reference
llm-essentials:agents/readmeThe summary Claude sees when deciding whether to delegate to this agent
<div align = "center"> [](python/AGENT.md) </div> <div align = "justify"> Agents are **_task-scoped specialists_** that the orchestrator (Claude Code, CodeX, Cursor AI, etc.) delegates work during a *multi-step* code change. Where a *skill* teaches the LLM **how** to format or shape c...
Agents are task-scoped specialists that the orchestrator (Claude Code, CodeX, Cursor AI, etc.) delegates work during a
multi-step code change. Where a skill teaches the LLM how to format or shape code, an agent teaches the LLM
when to think, how to think and with which discipline. Each agent ships as a directory (named after the programming
language or end-usage applcation) with an AGENT.md file with detailed orchestrations (in case sub-agents are involved)
and frontmatter that details the default model, etc. standard to an agent builder.
npx claudepluginhub pyutility/polyskills --plugin llm-essentialsExpert Go code reviewer that analyzes diffs, runs go vet and staticcheck, and checks for idiomatic Go, concurrency bugs, error handling, and security issues.