ChatGPT Plugin

CentricMem

Agent Memory Organising Layer

Brand: CentricMem

Category: AI
Primary Subcategory: Personal AI Memory & Second Brain

Integration details

Description

When you switch from Codex to Hermes, or from your laptop to another machine, the project should not start from zero. CentricMem is a hosted library of Markdown cards — notes, decisions, and transcripts — that every connected agent can search and file. File a decision in Codex today. Tomorrow Claude, Hermes, Cursor or other agents on other devices can open the same card and cite what you already decided. Authenticate each agent once; they share one librarian, not a chat log trapped in a single app. Share a named shelf with a teammate by email. Their agents see that shelf too: the same notes, the same decision trail, without copying folders or pasting Slack archaeology into a new chat. You keep an audit trail instead of tribal knowledge in scrollback. Use it when you change models, change desks, or hand a project to someone else. The cards are Markdown you can read. Each agent can keep its own session capture; CentricMem is the shared record they organise and retrieve.

Integration type: Plugin
Verification status: Not applicable
Platform: ChatGPT
Category: AI
Primary Subcategory: Personal AI Memory & Second Brain
Secondary Subcategories: None listed
Brand: CentricMem
Access: Account required
First tracked: 2026-09-19
Tool count: 17
Market: US

Agent DiscoverabilityChatGPT Plugin Discovery Score

ChatGPT organic discovery is live

Your score is coming

ChatGPT now suggests Plugins on its own when they match a user's request.Your Plugin Discovery Score measures how often yours appears, and it will show here as soon as it’s ready.

ChatGPT recommends Airtable when a user asks it to update a CRM. — ChatGPT surfaces Airtable because it matches the user's request.

Primary Subcategory

Competing in ChatGPT Personal AI Memory & Second Brain

View Category

Listed in alphabetical order

Primary Subcategory

2MindComing soonGet notified

AI PassportComing soonGet notified

AI Whisper Voice Note TakerComing soonGet notified

AI WisebaseComing soonGet notified

AiAkiv MemoryComing soonGet notified

CentricMemComing soonGet notified

Tools

17 tools agents can invoke

Session-start memory block. Does not write cards or .ambient.md. Pass shelf, library, or cwd so later writes route correctly. Unmatched cwd is not a shelf (library=none plus libraries=). Curate: empty means offer once to file existing durable memories as cards (capture stays).
cm_ambient
Copy shelf cards+attach on the librarian (from→to). Extra keys need both grants. Never to=unclassified. See Skill REFERENCE.
cm_copy
Delete leftover shelf {id}, or card {file,shelf} (optional heading= for one ##). Default key. Not account delete. See Skill REFERENCE.
cm_delete
Distinguish missing CLI, sandbox, librarian down, and token failure. Does not write memory cards.
cm_doctor
File a session-unit card on a named shelf. This does not close ChatGPT or the MCP session. summary= is required and must state the key points of this unit, not a placeholder. Pass attach= from keep. Skip if the user said not to remember.
cm_done
Host-specific operational detail, served on request: how each agent installs and refreshes this Skill, the failure modes worth knowing, the OAuth callback shapes, and how the optional hooks layer. Read it when a host-specific question blocks you - it is served here instead of shipping in every Skill copy.
cm_ops
Daily cards: items= (or package.items). A bundle that is {items:[...]} with no version is that same package ingest. ImportBundle is {version:1, lessons?, imported?, decisions?, sessions?, research?} — not a card list. skipExisting skips a dest slug (or mapped external_id) instead of writing slug-2.md. One commit max 50 cards / 50 attachments. Each item needs title (summary) and body (key points). Stage locally, cm_keep {card:false}, then items with attach pointers. Do not send a zip through MCP. dryRun previews the same skip/paths as apply.
cm_import
Store an original in R2 (prefer contentBase64 for binary). Never path= or secrets. Keep stub ≠ card — follow with cm_note. card=false for later cm_import. See Skill REFERENCE.
cm_keep
Check that the CentricMem librarian is reachable.
cm_health
Omit id to list shelves this key can see ({ok, libraries:[{id, displayName}]}). Default lists the catalog including share:; extra keys list grants only. With id, mint a named shelf (or rename the display label of an existing id). Extra keys can list; they cannot mint. Never id=unclassified. New ids past the named-shelf cap (Lite/Education 1, Pro 10, Ultra 50, Lifetime 1 20, Lifetime 2 100) return 403 SHELF_LIMIT.
cm_library
Write an architecture / host decision. title is the summary; decision/context/consequences are the key points. refs is a string: 1, 0001, #0001, or a comma list. File this turn when the choice is durable.
cm_log_decision
Move selected Markdown card paths between named shelves. files= whole paths only (not lessons.md / #heading). Default key. Never to=unclassified. See Skill REFERENCE.
cm_move
Rebuild FTS when the index may be dirty or after bulk/local-junction sync. Host cm_import already indexes. scanned = candidates, indexed = rehashed this pass, skipped = walked but not FTS. With no shelf= and a * / owner key, rebuilds every granted shelf (same as all=true). all=true is ignored when the key is pinned to one shelf.
cm_index
Rename card title {file,shelf,title}; optional heading= for multi-## files. Default key. See Skill REFERENCE.
cm_rename
FTS search (`q`, not `query`), or list cards when q is omitted and shelf= is set (browse default 50). Pass limit= to raise. tags/type/status/filter/shelf. See Skill REFERENCE.
cm_search
Print the Markdown card. Originals are human downloads, not agent context. Extra keys and * keys must pass shelf= / library= (or project=); omitting it returns Pick a shelf.
cm_show
Write durable knowledge. title is the summary; body is the key points later agents cm_show. Title-only is not a card. Same title in lessons.md is an error — pick a new title, or cm_delete heading= then rewrite. File as soon as Non-Micro work exists; never library=unclassified.
cm_note

FAQs

CentricMem ChatGPT Plugin FAQ

How the directory, categories and Discoverability Score work.

Read the methodology

How do I improve CentricMem's ChatGPT Plugin discoverability?

The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.

What are CentricMem alternatives on ChatGPT?

As of 2026-11-02, CentricMem competes with 2Mind, AI Passport, AI Whisper Voice Note Taker, AI Wisebase, AiAkiv Memory, AideMemo, Areta, AristAI Thinker and 114 more in ChatGPT Personal AI Memory & Second Brain, ranked by public Discoverability Score.

Where is this profile measured?

This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.