Issue Station
Sign In
Tools hereToolbox

hermes - 💡(How to fix) Fix Feature: Per-conversation context isolation for multi-chat deployments [1 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
NousResearch/hermes-agent#14162•Fetched 2026-04-23 07:46:29
View on GitHub
Comments
0
Participants
1
Timeline
4
Reactions
0
Author
thomasmarcelthomasmarcel
Participants
thomasmarcelthomasmarcel
Timeline (top)
labeled ×4

Currently, Hermes Agent's memory and state are global across all conversations/channels. When deployed to multiple chats (e.g., personal 1v1 + work group), there's no isolation between conversations — content from one chat could potentially leak to another.

This feature request is for proper per-conversation context isolation.

Error Message

  • Manual memory prefixing (error-prone, requires user discipline)

Root Cause

Currently, Hermes Agent's memory and state are global across all conversations/channels. When deployed to multiple chats (e.g., personal 1v1 + work group), there's no isolation between conversations — content from one chat could potentially leak to another.

This feature request is for proper per-conversation context isolation.

RAW_BUFFERClick to expand / collapse

Description

Currently, Hermes Agent's memory and state are global across all conversations/channels. When deployed to multiple chats (e.g., personal 1v1 + work group), there's no isolation between conversations — content from one chat could potentially leak to another.

This feature request is for proper per-conversation context isolation.

Use Case

  • User has personal 1v1 chat (Telegram) and work group chat
  • Personal conversations should stay personal
  • Work group discussions should stay in work context
  • No cross-contamination of memories, preferences, or context

Proposed Solution

  1. Namespace-based memory keys — prefix memory entries by conversation ID (e.g., work:user_preference, personal:memo)
  2. Conversation context tracking — know which "mode" the agent is operating in per session
  3. Isolation boundaries — explicit rules about what can/cannot be shared between contexts
  4. Optional: skill scoping — load different skills per conversation type

Benefits

  • Privacy-preserving across chat platforms
  • Clean separation for work vs. personal use cases
  • No risk of leaking personal info to work groups (or vice versa)
  • Better multi-chatbot deployments

Alternatives Considered

  • Using separate bot instances per conversation (wasteful, expensive)
  • Manual memory prefixing (error-prone, requires user discipline)

Additional Context

See discussion in https://github.com/NousResearch/hermes-agent/discussions (if any)

Tags: feature-request, memory, isolation, multi-chat

extent analysis

TL;DR

Implement namespace-based memory keys and conversation context tracking to achieve per-conversation context isolation.

Guidance

  • Introduce a conversation ID-based prefix for memory entries to prevent cross-contamination of memories and preferences.
  • Develop a mechanism to track the conversation context per session, enabling the agent to operate in the correct "mode".
  • Establish explicit rules for what can be shared between contexts to maintain isolation boundaries.
  • Consider implementing skill scoping to load different skills per conversation type for enhanced flexibility.

Example

# Example of namespace-based memory key
memory_key = f"{conversation_id}:user_preference"

Notes

The proposed solution requires careful consideration of the conversation ID generation and management to ensure unique and consistent identifiers across different chats and platforms.

Recommendation

Apply the proposed solution of namespace-based memory keys and conversation context tracking, as it provides a robust and scalable approach to achieving per-conversation context isolation without the need for separate bot instances or manual memory prefixing.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#memory management#API rate limit#retriever error#indexing error#inference speed

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING