{
  "slug": "hermes-agent",
  "name": "Hermes Agent",
  "category": "Harnesses / Personal Agents",
  "tagline": "Nous Research's self-improving agent: skills, cross-session memory and a messaging gateway, pointed at your local model.",
  "kit_says": "Most harnesses forget you when the session ends. Hermes is the one trying to learn, which is exactly the part of the gap local setups usually skip.",
  "homepage": "https://hermes-agent.nousresearch.com/",
  "repo": "https://github.com/NousResearch/hermes-agent",
  "docs": "https://hermes-agent.nousresearch.com/docs/",
  "license": "MIT",
  "platforms": [
    "macos-apple-silicon",
    "linux",
    "windows"
  ],
  "mac_studio_fit": "Strong",
  "what_it_is": "A general-purpose agent with a terminal UI and a gateway for Telegram, Discord, Slack, WhatsApp, Signal and Email. It creates skills from experience, keeps agent-curated memory, searches past sessions, and runs scheduled automations. Works with any OpenAI-compatible endpoint, including Ollama.",
  "gap_role": "Harness + memory: a built-in learning loop (skills and cross-session recall) is one of the frontier-app features a bare local model lacks.",
  "prerequisites": [
    "A local OpenAI-compatible server (Ollama at http://127.0.0.1:11434/v1)",
    "A model with tool calling and a large context window (Ollama recommends at least 64000 tokens for agents)"
  ],
  "install": {
    "macos-apple-silicon": [
      {
        "run": "curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash"
      },
      {
        "run": "source ~/.zshrc",
        "note": "reload your shell so `hermes` is on PATH"
      }
    ],
    "linux": [
      {
        "run": "curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash && source ~/.bashrc"
      }
    ],
    "windows": [
      {
        "run": "iex (irm https://hermes-agent.nousresearch.com/install.ps1)",
        "note": "PowerShell"
      }
    ]
  },
  "configure": [
    {
      "run": "ollama launch hermes",
      "note": "easiest path: Ollama installs Hermes if needed, points it at http://127.0.0.1:11434/v1 and sets your model as primary"
    },
    {
      "run": "hermes model",
      "note": "manual path: choose More providers → Custom endpoint → http://127.0.0.1:11434/v1, leave API key blank"
    },
    {
      "run": "OLLAMA_CONTEXT_LENGTH=64000 ollama serve",
      "note": "Ollama docs: agents need at least 64000 tokens of context"
    },
    {
      "run": "hermes gateway setup",
      "note": "optional: chat with it from Telegram/Discord/Slack/WhatsApp/Signal/Email"
    }
  ],
  "verify": [
    {
      "run": "hermes doctor",
      "expect": "diagnostics report with no blocking errors"
    },
    {
      "run": "hermes",
      "expect": "TUI opens; a one-line prompt gets an answer from the local model"
    }
  ],
  "rollback": [
    {
      "run": "hermes uninstall --dry-run",
      "note": "preview what will be removed"
    },
    {
      "run": "hermes uninstall",
      "note": "default keeps config and data in ~/.hermes; add --full to remove data too (destructive, run `hermes backup` first)"
    }
  ],
  "caveats": [
    "Large, fast-moving project (multiple pushes per day); commands can change, re-check the docs.",
    "Nous Portal is an optional paid provider; this listing covers the local Ollama path only.",
    "Not yet run on the Mac Studio test rig."
  ],
  "works_with": [
    "ollama",
    "lm-studio",
    "litellm",
    "mcp-servers",
    "agents-md"
  ],
  "sources": [
    "https://github.com/NousResearch/hermes-agent",
    "https://hermes-agent.nousresearch.com/docs/getting-started/installation",
    "https://hermes-agent.nousresearch.com/docs/getting-started/updating",
    "https://docs.ollama.com/integrations/hermes",
    "https://docs.ollama.com/context-length"
  ],
  "sponsored": false,
  "verified_on": "2026-10-15",
  "gap_score": {
    "value": null,
    "status": "pending test on Mac Studio",
    "baseline": "frontier reference (see /about#gap-score)"
  },
  "benchmarks": {
    "status": "pending test on Mac Studio",
    "rig": "Mac Studio, Apple Silicon, 128GB unified memory",
    "results": []
  },
  "hardware_tested": [],
  "repo_meta": {
    "stars": 254039,
    "stars_as_of": "2026-10-15",
    "last_push": "2026-10-15",
    "archived": false,
    "license_spdx": "MIT"
  },
  "stack_layer": "harness",
  "summary": {
    "what": "A personal agent that learns skills and remembers you.",
    "why": "Your local model gets a learning loop, not just a chat box."
  }
}