Connect Hermes Agent to Paperclip so AI agents automatically execute tasks assigned through Paperclip's employee-style interface.
Set up an AI agent that wakes up and processes a Paperclip task whenever a team member leaves a comment, without manual intervention.
Run multi-step file editing, web browsing, and terminal tasks through Hermes inside Paperclip with session state preserved across restarts.
Switch the underlying AI model between Anthropic, OpenAI, OpenRouter, or five other providers without changing your Paperclip workflow.
Requires both npm (for the adapter) and pip (for Hermes Agent) to be installed and configured before use.
This repository is a connector between two AI-related platforms: Paperclip and Hermes Agent. Paperclip is a service for managing AI agents as if they were employees in a company, assigning them tasks and tracking their work. Hermes Agent is a separate AI agent developed by Nous Research that comes with over 30 built-in tools (for browsing the web, working with files, running terminal commands, and more), persistent memory across sessions, and support for multiple AI model providers. The adapter makes it possible to run a Hermes Agent inside Paperclip's environment. Without this connector, the two systems would not know how to talk to each other. With it, you can create an agent in Paperclip, assign it a task, and have Hermes Agent do the actual work, reporting results back to Paperclip when finished. The agent can also wake up when someone leaves a comment on a task, not just when a task is first assigned. The technical work the adapter does includes translating Hermes's raw text output into structured entries that Paperclip can display with proper formatting, cleaning up Hermes's ASCII-style banners and table borders into standard markdown, and keeping session state so that each run picks up where the last one left off. It also handles cases where Hermes logs messages to the error output that are not actually errors, so they do not show up as failures in Paperclip's interface. Configuration covers which AI model and provider to use (eight providers are supported, including Anthropic, OpenAI, and OpenRouter), how long to let the agent run before stopping it, whether to save filesystem snapshots for rollback, and which toolsets to give the agent access to. Installation is through npm, and Hermes Agent must be installed separately via pip before the adapter can function. The project is written in TypeScript.
← williambryk on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.