Product updates

Your dashboard now shows your account limits at a glance — capsules, memories, tags, API keys, and rate limit — so you always know where you stand.

Rate limiting is now per account. All your API keys and capsules share a single rate-limit bucket, giving you a consistent and predictable quota across your entire account.

You can now read and write capsule memories using plain HTTP requests. The new REST API covers every operation available through MCP — store, retrieve, delete, search, bulk operations, tags, and capsule info — with familiar REST conventions.

This is useful when you want to:

• Integrate agimem into backends, scripts, or pipelines that don't use an MCP client.
• Build custom dashboards or admin tools on top of your capsule data.
• Quickly test or debug memories with curl or any HTTP client.

Authentication

Use the same API key you already have. Pass it as a Bearer token:

Authorization: Bearer mcp_...

All endpoints live under /api/rest/. Rate limits, memory limits, and tag limits are the same as the MCP interface.

See the full REST API documentation for endpoints, request/response examples, and details.

You can now clone a capsule directly from the dashboard. Hit the copy icon on any capsule card, give it a new name, and you'll get an exact duplicate with all non-expired memories, tags, and memory-tag associations carried over.

This is useful when you want to:

• Fork a capsule to experiment without affecting the original.
• Set up a new agent with the same baseline context as an existing one.
• Create environment-specific copies (e.g. staging vs. production conventions).

Agents can now work with capsule memory more efficiently with four new tools:

• memory_count — returns the number of stored memories and the capsule limit, so agents know how much room is left without listing everything.
• memory_keys — lists only key names with no values attached. Perfect for orientation at the start of a session.
• memory_exists — a fast boolean check for a single key, without pulling its value or tags.
• memory_list_by_tags — filters keys by multiple tags at once (AND logic). Find memories that match all the categories you care about in a single call.

These tools reduce token usage and round trips for the most common "what do I have?" queries agents run when starting a new conversation.

Capsule Info

Also new: capsule_info gives agents a quick snapshot of the current capsule — name, memory count vs limit, tag count vs limit, and last activity. One call to orient before doing any work.

Each capsule on your dashboard now has a small status dot next to its name:

• Green and pulsing — your agent is actively using this capsule right now.
• Muted — idle. Hover for the last activity time, or "No agent has connected yet" if it hasn't been used.

Paste your API key into Cursor, Claude Desktop, or VS Code, switch back to the dashboard, and watch the dot turn green the moment your agent makes its first call. No more guessing whether things are wired up correctly.

Two new MCP tools complete the bulk memory workflow alongside memory_bulk_set:

• memory_bulk_get — retrieve up to 50 keys in one call. Each result tells the agent whether the key was found, along with its value and tags.
• memory_bulk_delete — remove up to 50 keys in one call, with a clear per-key status in the response.

Replacing a loop of memory_get or memory_delete calls with a single bulk request means fewer tool invocations, lower latency, and noticeably fewer tokens spent on protocol overhead — leaving more of the context window for the actual work.

You can now get in touch with us quickly and easily for any questions, feature requests, or support inquiries. The new Contact & Support page lets you send messages straight to the team.

Fill in your name, email, subject, and message — we'll get back to you as soon as we can.

Agents can now store multiple memories at once using the new memory_bulk_set MCP tool. Pass an array of up to 20 items, each with a key, value, and optional tags and ttlMinutes — all written in a single call instead of repeated memory_set requests, and returns a per-key status summary so agents know exactly what succeeded.

Agents can now search across all memories in a capsule using the new memory_search MCP tool. It performs a case-insensitive search across both keys and values, returning matching results with a preview of each value.

Pass a query string to find what you need, and optionally set a limit (1–50, default 20) to control how many results are returned. Expired memories are automatically excluded from results.

Memories can now expire automatically. When storing a memory, set a TTL (1 minute to 1 year) and it will be automatically removed once that time elapses — no manual cleanup needed.

Memories without a TTL remain permanent, so existing behavior is unchanged. The dashboard also shows an expiration time for any memory with a TTL set.

Tags are now a first-class feature in agimem, enabling powerful memory organization and intelligent agent context retrieval.

What You Get

• Tags Tab: View all available tags and their usage across each capsule.
• Inline Tags: Tags appear directly on memory rows for easy classification reference.
• MCP Support: Agents can now query memories by tag to retrieve relevant context automatically.

Use It

In the UI: Open the Tags tab to see how memories are organized and classified.

In Agents: Full tag management through the MCP API. Agents can create, list, update, and delete tags, as well as query memories by tag to retrieve filtered results. Build dynamic taxonomies and intelligent context retrieval for context-aware decision making and knowledge organization.

We've launched a dedicated changelog to keep you informed about what's new in agimem. All product updates, improvements, and releases will be published here for easy reference.

The changelog is a living record of our product development — visit it regularly to stay up to date with the latest features and enhancements.