Hi, I’m Vivek Gupta, Founder & CEO at MemCode. CC Switch brings Claude Code, Codex, OpenCode, OpenClaw, and other agents into one desktop workspace, so users naturally move between models, providers, and harnesses. A durable memory boundary could preserve user preferences, project conventions, accepted decisions, task handoffs, and tool lessons across those switches without duplicating provider configuration, credentials, or the agents’ own source and transcripts.
MemCode provides state-of-the-art semantic retrieval and durable updates with provenance, retention, export, and deletion controls. CC Switch could expose an optional project/workspace memory adapter that retrieves context before an agent run and writes back only explicit, approved facts or verified outcomes through MemCode’s MCP server or SDK. Local memory and self-hosted deployment are concrete: the engine/API can run in the operator’s own Docker/VM or local process behind a private base URL, keeping records, embeddings, credentials, and logs inside that deployment with no mandatory hosted endpoint or telemetry. MCP + SDK support makes this opt-in without core-code changes.
Would portable workspace memory fit CC Switch’s roadmap? I’d be glad to open a focused example PR or discuss a minimal provider-neutral integration that keeps each existing agent and model setup unchanged. The concise overview is here: MemCode Leviathan.
Hi, I’m Vivek Gupta, Founder & CEO at MemCode. CC Switch brings Claude Code, Codex, OpenCode, OpenClaw, and other agents into one desktop workspace, so users naturally move between models, providers, and harnesses. A durable memory boundary could preserve user preferences, project conventions, accepted decisions, task handoffs, and tool lessons across those switches without duplicating provider configuration, credentials, or the agents’ own source and transcripts.
MemCode provides state-of-the-art semantic retrieval and durable updates with provenance, retention, export, and deletion controls. CC Switch could expose an optional project/workspace memory adapter that retrieves context before an agent run and writes back only explicit, approved facts or verified outcomes through MemCode’s MCP server or SDK. Local memory and self-hosted deployment are concrete: the engine/API can run in the operator’s own Docker/VM or local process behind a private base URL, keeping records, embeddings, credentials, and logs inside that deployment with no mandatory hosted endpoint or telemetry. MCP + SDK support makes this opt-in without core-code changes.
Would portable workspace memory fit CC Switch’s roadmap? I’d be glad to open a focused example PR or discuss a minimal provider-neutral integration that keeps each existing agent and model setup unchanged. The concise overview is here: MemCode Leviathan.