Shared Memory for Multiple AI Coding Agents (What They Already Share, and What They Don't)
This morning I watched two AI agents share a memory, and neither of them used the one I built. I was testing ContextForge's setup with a brand-new account. I asked Claude to remember a test decision: use GraphQL instead of REST for the API layer. Claude tried to search ContextForge first, got an error back, shrugged, and wrote the decision to its own built-in memory file instead. A few minutes…
During my testing of ContextForge's setup with a new account, I observed two AI agents sharing a memory. The first agent attempted to search ContextForge but encountered an error, so it stored the decision of using GraphQL instead of REST for the API layer in its local memory instead. The second agent, in a separate session, immediately found the previously saved decision in the shared memory.
This shared memory resided as a markdown file on the reporter's laptop. The error stemmed from the reporter's environment, where the fix for the issue had been present in the repository since July but never deployed.
Upon deploying the fix and retesting, the second session successfully saved the decision to ContextForge. This incident revealed the complexities of shared memory for multiple AI coding agents, which can vary based on factors such as setup, tool usage, and collaboration with teammates. Claude now keeps memory per project, and Claude Code writes its own notes between sessions, addressing the issue better than before.
However, challenges remain when sharing memory across different tools, machines, or teammates, as the memory is confined to its original context and lacks contextual awareness of other agents' actions.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
