Engram

Store messages in Engram memory, verbatim and automatically chunked + embedded for search. conversation_id is OPTIONAL: omit it to append to the user's default memory (recommended for general 'remember this' requests) — never ask the user for an id. Pass a conversation_id (from create_conversation) only when you want to group a specific topic. The response returns the conversation_id used. Optionally accepts client-encrypted vault entries for secrets detected client-side.

append_messages

Create a new conversation and return its conversation_id. Call this yourself to obtain an id before appending — the id is yours to generate and reuse; never ask the user to provide one. Create one conversation per session/topic and reuse its id for all subsequent append_messages calls.

create_conversation

Delete a conversation and all its messages, chunks, and vector embeddings.

delete_conversation

Get a conversation with its full verbatim messages. Supports pagination.

get_conversation

List conversations with filtering and sorting options.

list_conversations

Show how full the user's Engram memory is: lifetime storage used vs the plan's capacity, plus a monthly meter on plans that have one. Returns a ready-to-display progress bar string.

memory_status

Searches the user's long-term memory. Applicable when previously saved context is relevant to the current request — for example when the user references something from before ('what did we decide', 'like last time', 'my usual', a project/person/preference by name) or asks what is remembered about a topic. Hybrid search (semantic + keyword) across stored conversations. Each result includes a short chunk_summary, the full chunk_text snippet, conversation_title, tags, and relevance scores, best match first.

search

Show which Engram account this connection is using — organization name, email, tier, and how you are authenticated. Use when unsure which account memories are being saved to or searched from.

whoami