HN Showcase
Back to browse
Termem, cross-agent memory and session management

Termem, cross-agent memory and session management

by todience·Jun 15, 2026·2 points·0 comments
Visit ProjectView on HN

AI Analysis

●●●BangerZero to OneBig BrainNiche Gem

Claude can read Codex sessions—cross-agent memory without network calls.

Strengths
  • •Cross-agent memory sharing is genuinely novel—different agents see each other's work
  • •Zero network calls, all local storage for privacy and speed
  • •MCP server integration enables agents to recall prior sessions automatically
Weaknesses
  • •Requires agents to support MCP for full functionality
  • •Early stage with only 45 commits, unproven at scale
Target Audience

Developers using multiple AI coding agents in terminal workflows

Post Description

I built Termem (terminal memory) to allow quick agent session recovery, and cross-agent memory sharing. Once I cd into a folder, I can run 'termem' to view all agent sessions and resume them. When I launch a fresh session, the agent can recall all sessions within that directory via the mcp & skill. Claude can see sessions by codex and other agents.

Lemme know your thoughts.

Similar Projects

AI/ML●Mid

Residuum | Agentic AI with continuous context

OpenClaw fork with continuous memory instead of RAG — solves a real pain, not groundbreaking.

Ship It
BearFlinn
103mo ago