Repolode
Submit
42-evey

42-evey/hermes-plugins

A collection of 23 custom plugins for hermes-agent covering autonomy, model routing, observability, cost control, memory and inter-agent messaging.

View on GitHubWebsite
Stars
532
Forks
60
Open issues
0
Last commit
3 months ago

What is it?

This repository is a set of 23 plugins for hermes-agent, an autonomous AI agent framework. Each plugin is a separate directory that is copied into the agent's plugins folder, and they can be installed selectively. The plugins are grouped into autonomy and decision-making, observability and telemetry, quality and safety, learning and memory, and communication and integration.

The plugins address operational concerns for long-running agents: budget enforcement through Langfuse with warnings at 80% and alerts at 100%, model routing with retries and a fallback chain, prompt injection screening for incoming email, and structured JSON logging of tool calls. Others handle goal tracking, memory scoring and consolidation into MEMORY.md and Qdrant, and a file-based plus MCP bridge to Claude Code.

A shared evey_utils.py module provides retry logic with exponential backoff and HTTP helpers used by several plugins. The project is MIT licensed and targets users already running hermes-agent.

Key features

  • 01Budget enforcement via Langfuse with warnings at 80% and alerts at 100%.
  • 02Model routing with per-model retries and a four-model fallback chain.
  • 03Structured JSON telemetry of tool calls, delegations and errors.
  • 04Prompt injection screening combining regex patterns and a local classifier.
  • 05Goal management tools to set, track, complete and review goals.
  • 06Bidirectional bridge with Claude Code using file inboxes and an MCP server.

Use cases

  • Running a hermes-agent instance continuously while keeping model API spend under a daily budget.
  • Routing sensitive tasks to local models while sending other work to cloud models.
  • Connecting a hermes-agent to Claude Code so the two agents can exchange messages.
  • Screening inbound email before an agent acts on its content.

Topics

#ai-agent#plugins#llm#observability#cost-control#model-routing#memory#autonomous-agent#hermes-agent#open-source

FAQ

What is hermes-plugins?
A collection of 23 custom plugins for hermes-agent covering autonomy, model routing, observability, cost control, memory and inter-agent messaging. This repository is a set of 23 plugins for hermes-agent, an autonomous AI agent framework. Each plugin is a separate directory that is copied into the agent's plugins folder, and they can be installed selectively. The plugins are grouped into autonomy and decision-making, observability and telemetry, quality and safety, learning and memory, and communication and integration.
Is hermes-plugins free to use? What license does it have?
hermes-plugins is open source under the MIT license. Check the license terms before using it in commercial products.
Is hermes-plugins still maintained?
Updated within 6 months. The last commit to hermes-plugins was on Jul 18, 2026.
What can I use hermes-plugins for?
Running a hermes-agent instance continuously while keeping model API spend under a daily budget. Routing sensitive tasks to local models while sending other work to cloud models. Connecting a hermes-agent to Claude Code so the two agents can exchange messages. Screening inbound email before an agent acts on its content.
Where can I get hermes-plugins?
The source code and installation instructions are on GitHub: https://github.com/42-evey/hermes-plugins

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects