Journalism agent skills A collection of Agent Skills for journalists, researchers, academics, media professionals, and communications practitioners. The same repository serves Claude Code and Codex while keeping Claude only commands, agents, and hooks clearly labeled. Docs site: skills.amditis.tech — interactive skill browser, setup guides, and full documentation. Guides Setup and workflow guides, separate from the skills themselves: Guide Description Autonomous and hands on dev work A first person account of running dev work with Claude in two modes — autonomous sessions that pull tasks off GitHub issues, and hands on multi agent reviews at the keyboard — held to one quality bar, with copy able prompts to adapt the approach to your own agent Multi agent workflows Plain language guide to running many AI agents on one job — fan out, pipeline, and adversarial verify patterns, with real examples from ICIJ, The Markup, Full Fact, and Elicit, plus honest cautions Persistent sessions Keep Claude Code sessions alive through disconnects using tmux — setup, key bindings, activity notifications, and scheduler coexistence What are agent skills? Skills are modular instruction sets that an AI coding agent loads when relevant to a task. Each skill contains domain knowledge, workflows, templates, and supporting files. Shared skills in this repository follow the Agent Skills specification; Claude and Codex provide their own installation and runtime layers around that shared content. Installation Choose the section for your client. During the Codex pilot, use only one Codex installation path for a given skill or package. Claude Code Prerequisite: You need Claude Code installed. Run claude version in your terminal to check. Plugins (recommended) Plugins give you slash commands you can run directly inside Claude Code. Run these two commands inside a Claude Code session: [安装/使用命令见下方] Then restart Claude Code (close and reopen). See the PDF Playground README for detailed setup instructi