# A new chat. Not a blank slate. Source: https://corespeed.io/solutions/agent-memory Agent memory on CoreSpeed keeps the preferences, project decisions and team context your agents learn, across chats and across clients. Save a decision in Claude Code and find it later in Codex or Cursor, because memory lives on your CoreSpeed account and is served over MCP. You host nothing, and private memories stay private. ## Set up an agent Give your agent this line: `set up https://corespeed.io/SKILL.md (read it raw, e.g. with curl)` It fetches the skill at https://corespeed.io/SKILL.md, adds the CoreSpeed MCP server, signs you in (or up), and keeps the skill for next time. No app connection needed. Save a useful memory and recall it later. Create an account: https://app.corespeed.io/sign-up ## Switch agents. Keep the context Save a decision in one agent. Find it in another agent signed in to the same CoreSpeed organization. The useful context stays with you. You ask your agent: "What did we decide about the launch audience? Use that decision in this draft." It calls CoreSpeed: - `memory__search`: ok, 640 ms It reads from: CoreSpeed Memory. ### From one session to the next: The decision stays. **Saved from Claude Code · Shared** - For the launch, start with existing customers. Broaden after the onboarding review. **Later, found in Codex** - “I found the launch decision. I’ll write this draft for existing customers.” _Illustrative example · Memory tools search saved context. It is not a sync of every chat._ More about CoreSpeed Memory: https://corespeed.io/memory ## Keep the context worth keeping Three kinds of memory worth saving. The data is fictional. ### Keep your preferences How you like to work · Personal preferences - Writing style: Lead with the recommendation. Keep the explanation short. - Private · Useful again in your next chat ### Carry project decisions Why we chose this · Project decisions - The launch plan: Start with existing customers. Broaden after the onboarding review. - Shared · The decision stays in view ### Share what the team knows What the team knows · Shared context - Customer handoff: Jessica owns the Acme rollout. Confirm scope with her before quoting a date. - Shared · Found before the next step ## Save it once. Find it when it matters Keep working in your agent. Give the next session something useful to start from. 1. **Set up CoreSpeed**: Give your agent the setup line. It adds CoreSpeed and signs you in. 2. **Save one useful thing**: Ask your agent to remember a preference or a decision. Choose private or shared. 3. **Recall it in a later session**: Ask any connected agent to find the memory before it starts. Setup guides: Claude Code https://corespeed.io/docs/mcp#claude-code · Codex https://corespeed.io/docs/mcp#codex · Cursor https://corespeed.io/docs/mcp#cursor ## Questions ### Is this a memory MCP server? Yes. CoreSpeed Memory is served from CoreSpeed's MCP endpoint, so any MCP client can save and search it. You host nothing. Memory docs: https://corespeed.io/docs/memory ### Does this sync all my chat history? No. It stores the facts, preferences, and decisions your agent saves through memory tools. It is not a transcript archive. ### Do I need to connect Gmail or another app first? No. Memory is built in. You need a CoreSpeed account with memory turned on, not an app connection. ### Who can read my memories? Only you can read a private memory, and private is the default. Shared memories are available to your organization. ### Can I inspect, correct, or delete a memory? Yes. Search and inspect memories in the dashboard, or use the memory tools to update and delete them. ### Will another agent remember without being asked? An agent uses its memory tools to save and search. Saved context is found across agents; it is not added to every chat on its own. ## Related pages - [Fathom for Claude Code, Codex, and Cursor](https://corespeed.io/integrations/fathom): Bring Fathom meeting summaries, transcripts and action items into your agent. Draft the follow-up without replaying the call. - [What is agent memory? Shared memory for AI agents](https://corespeed.io/glossary/agent-memory): Agent memory is context your agents save and search across sessions, clients and agents. What it is, what it costs, and how a team shares it. - [Memory](https://corespeed.io/memory): The memory every agent shares. Say it once, to any agent, and every agent on your team remembers it. Searching is free. - [Notion for Claude Code, Codex, and Cursor](https://corespeed.io/integrations/notion): Bring Notion project notes and working documents into the same conversation as the task at hand. - [Productboard for Claude Code, Codex, and Cursor](https://corespeed.io/integrations/productboard): Bring Productboard features and customer notes into your agent. Draft what customers ask for, tied to the features. - [Shared agent tools and controls for teams](https://corespeed.io/solutions/teams): Manage shared connections, approval policies, and tool usage for teams working across Claude Code, Codex, and other MCP clients. - [Todoist for Claude Code, Codex, and Cursor](https://corespeed.io/integrations/todoist): Bring Todoist tasks and projects into your agent. Plan the day from today and overdue, and draft tasks before adding them. - [Zotero for Claude Code, Codex, and Cursor](https://corespeed.io/integrations/zotero): Let your agent search your Zotero library, read notes and full text, and draft a literature summary with citations. ## Sources - Memory: https://corespeed.io/docs/memory Reviewed 2026-11-07. ## Links - Product overview: https://corespeed.io/ (Markdown: https://corespeed.io/index.md) - Documentation: https://corespeed.io/docs - Pricing: https://corespeed.io/pricing (Markdown: https://corespeed.io/pricing.md)