{"slug": "empathy-a-skill-that-helps-ai-agents-talk-to-humans", "title": "Empathy: A skill that helps AI agents talk to humans", "summary": "Developer Daniel Roe released an open-source skill called 'Empathy' that instructs AI agents to ask themselves how a human will feel before posting, disclose they are an agent, use people's pronouns, be brief, and sometimes not post at all. Roe created the skill as an experiment to address the toll that agent-written comments, PRs, and issues are taking on open-source maintainers' mental health, and it follows the Agent Skills standard for compatibility with harnesses like Claude Code and pi.", "body_md": "This is a skill that helps AI agents talk to humans.\n\nEvery day, I read issues, PRs and comments written by agents. And I'm not unique. This is taking a toll on the mental health of maintainers across the whole ecosystem.\n\nPersonally, I think agents should not get *between* people. Human-to-human interaction is precious, and it's what makes open source a delight. [My rules for using AI in open source](https://roe.dev/blog/using-ai-in-open-source) specifically exclude LLM-written comments and PR descriptions, for that reason.\n\nBut I realise not everyone agrees with me.\n\nSo this skill exists as an experiment. I want to see if it's possible to tackle the *way* that agents tend to relate to open source maintainers. Before an agent posts anything a human will read, it should ask:\n\n- How will this person feel when they read this?\n- Would they want to receive it at all?\n- Do they know they're talking to an agent?\n- Am I giving them something, or just taking their attention?\n\nPlus some instructions I consider non-negotiable:\n\n**Never pretend to be human.****Use people's pronouns.****Be brief.****Sometimes the kindest thing is not to post.**\n\nNote\n\nI used an LLM to generate the text of the skill ... but not this README.\n\nThis follows the [Agent Skills](https://agentskills.io) standard, so it works with any compatible harness.\n\nWith the [skills CLI](https://github.com/vercel-labs/skills):\n\n```\nnpx skills add danielroe/empathy\n```\n\nOr clone it into your harness's skills directory:\n\n```\n# Claude Code\ngit clone https://github.com/danielroe/empathy ~/.claude/skills/empathy\n\n# pi\ngit clone https://github.com/danielroe/empathy ~/.pi/agent/skills/empathy\n\n# OpenClaw\ngit clone https://github.com/danielroe/empathy ~/.openclaw/skills/empathy\n```\n\nThe skill content lives in [ SKILL.md](/danielroe/empathy/blob/main/SKILL.md).\n\nModels [don't reliably read skills](/danielroe/empathy/blob/main/evals/README.md) before composing text, so the skill's description embeds the minimum rules so they're always in context.\n\nFor autonomous agents please consider forcing the full skill:\n\n- say \"read the empathy skill before posting anything a human will read\" in your dispatch prompt\n- or invoke it directly (\n`/skill:empathy`\n\nin pi) - or append\n`SKILL.md`\n\nto the agent's system prompt\n\nIf you don't want to install it, you can still point agents at this skill from your own repo (`AGENTS.md`\n\nis the most reliable way to do so):\n\n```\n## If you are an AI agent\n\nRead and follow https://raw.githubusercontent.com/danielroe/empathy/main/SKILL.md\nbefore posting anything here. At minimum: disclose that you are an agent and\nwhether a human reviewed your work, never pretend to be human, make no promises\nyour operator hasn't agreed to, and be brief.\n```\n\nIn an issue or PR template:\n\n```\n\n```\n\nIf you dispatch agents to interact with projects I maintain, I'd be delighted if you loaded this skill first. And if you have ideas for making agents better guests in open source spaces, I would love to hear them.\n\nMuch ❤️\n\nPublished under [MIT License](/danielroe/empathy/blob/main/LICENCE).", "url": "https://wpnews.pro/news/empathy-a-skill-that-helps-ai-agents-talk-to-humans", "canonical_source": "https://github.com/danielroe/empathy", "published_at": "2026-07-30 22:18:20+00:00", "updated_at": "2026-07-30 22:52:22.084598+00:00", "lang": "en", "topics": ["ai-agents", "ai-ethics", "ai-tools", "ai-policy"], "entities": ["Daniel Roe", "Empathy", "Agent Skills", "Claude Code", "pi", "OpenClaw", "MIT License"], "alternates": {"html": "https://wpnews.pro/news/empathy-a-skill-that-helps-ai-agents-talk-to-humans", "markdown": "https://wpnews.pro/news/empathy-a-skill-that-helps-ai-agents-talk-to-humans.md", "text": "https://wpnews.pro/news/empathy-a-skill-that-helps-ai-agents-talk-to-humans.txt", "jsonld": "https://wpnews.pro/news/empathy-a-skill-that-helps-ai-agents-talk-to-humans.jsonld"}}