honcho

Configure Honcho for Hermes cross-session memory and multi-profile isolation.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/MarbleSodas/Mavis --skill honcho-marblesodas
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: honcho
Source: https://github.com/MarbleSodas/Mavis/tree/main/optional-skills/autonomous-ai-agents/honcho
Command: npx skills add https://github.com/MarbleSodas/Mavis --skill honcho-marblesodas

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Honcho helps Hermes remember who the user is across sessions, keep multi-profile peers isolated, and reliably inject the right memory context without blowing the token budget.

Core Features & Use Cases

  • Cross-session user modeling: Builds a unified user representation over time for better continuity.
  • Multi-profile peer isolation: Gives each Hermes profile its own AI peer identity while sharing the user view.
  • Observation & recall control: Tunes what each peer learns from and chooses between hybrid/context/tools recall modes.
  • Dialectic reasoning configuration: Adjusts cadence, depth, and reasoning level for synthesized insights when needed.
  • Session strategy & summaries: Controls session scoping (per-directory/per-repo/per-session/global) and uses session summary injection.
  • Context budget enforcement & sanitization: Caps injected context tokens and sanitizes injected memory to reduce prompt-corruption edge cases.

Quick Start

Configure Honcho for Hermes by running the setup wizard and selecting cloud or self-hosted, then verify connectivity with the status command.

Frequently Asked Questions about honcho

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enable cross-session memory for Hermes deployments?

Configure Honcho with Hermes to enable cross-session memory and persistent personalization. This integration builds a unified user representation over time, ensuring continuity across directories, repositories, or sessions without losing context.

Can I isolate multi-profile peer identities while sharing a single user view?

Yes, multi-profile peer isolation gives each Hermes profile its own AI peer identity while sharing the unified user view. This prevents context bleed between distinct profiles during concurrent operations.

How do I stop injected memory context from exceeding my token budget?

Prevent token budget overflows by enabling context budget enforcement and sanitization. This caps injected context tokens and sanitizes memory content to reduce prompt-corruption edge cases during recall.

What recall modes are available when tuning Hermes observation and memory settings?

Hermes observation and recall control supports hybrid, context, and tools recall modes. These settings determine what each isolated peer learns from and how memory context is retrieved during sessions.

Does Honcho support dialectic reasoning configuration for synthesized insights?

Yes, Honcho supports dialectic reasoning configuration by adjusting cadence, depth, and reasoning level. This tunes synthesized insights for Hermes peers when deeper contextual analysis is required.

Do I need a self-hosted Honcho instance to use Hermes memory configuration?

No, you can use either cloud or self-hosted Honcho instances. Run the setup wizard to select your preferred deployment and verify connectivity with the status command before configuring Hermes.