hermes-memori 0.1.2

Official Memori Labs provider for Hermes Agent, enabling structured long-term memory from agent trace and execution.

Memori for Hermes Agent

Memori gives Hermes Agent structured, long-term memory from agent trace and execution. It captures completed agent activity — including user goals, assistant decisions, tool usage, workflow steps, outcomes, constraints, failures, and feedback — and structures that activity into durable memory primitives for future recall. This allows Hermes agents to learn from prior execution, preserve workflow context, avoid repeated mistakes, and become more efficient over time. The provider exposes explicit tools for memory recall, summaries, quota checks, signup, and feedback.

Requirements

• Hermes Agent with memory provider plugins
• A Memori API key
• Python 3.10+

Install

From this repository:

pip install -e .
pip install -e integrations/hermes

Or install the published package when available:

pip install hermes-memori

Configure

Use Hermes' memory setup flow and select memori:

hermes memory setup

If memori is not listed yet, install hermes-memori in the same Python environment Hermes uses, then set the provider manually.

Manual configuration also works:

hermes config set memory.provider memori
HERMES_HOME="${HERMES_HOME:-$HOME/.hermes}"
mkdir -p "$HERMES_HOME"
echo "MEMORI_API_KEY=your-key" >> "$HERMES_HOME/.env"

Then add $HERMES_HOME/memori.json:

{
  "entityId": "your-user-or-workspace-id",
  "projectId": "hermes"
}

Environment variables override file config:

• MEMORI_API_KEY
• MEMORI_ENTITY_ID
• MEMORI_PROJECT_ID
• MEMORI_PROCESS_ID

MEMORI_PROJECT_ID is optional. When it is not configured, Hermes-provided project context such as the active workspace, agent identity, user ID, session title, or session ID is used as the Memori project scope.

Tools

• memori_recall
• memori_recall_summary
• memori_quota
• memori_signup
• memori_feedback

Behavior

The provider is intentionally fail-soft. Memori network failures are logged but do not stop Hermes from answering the user.