ctxstore — memory that follows you across models MCP server
ai.ctxstore/ctxstore-mcpctxstore.aistreamable-httpv1.1.2Scanned
Persistent memory for AI agents: keyed facts, a wake bundle each session, one memory in every model.
33 tools: 3 can delete or cancel things, 3 do something that can't be undone, 15 change things and 12 only read.
- Reads private data · 1
- Reads text others write · 4
- Can send data out · 3
- 1 finding
Tools listed without sign-in. Remote posture
With Anlyon
3 tools here can delete or cancel things and 3 do something that can't be undone. Govern them.
- Anlyon lists the tools and adopts none.
- Each tool you adopt gets approvals and receipts.
- Your agent never holds the key.
A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.
What the ctxstore — memory that follows you across models MCP server can do
The ctxstore — memory that follows you across models MCP server is published by ctxstore.ai, a domain the MCP Registry verified. It has 33 tools. It can delete or cancel things (3 tools: delete_fact, leave_grant and revoke_grant), do things that cannot be undone, such as sending, paying or publishing (accept_invite, decline_invite and invite) and create or change data (15 tools: bind_agent, ghost_mode, index_stats, introduce_yourself and 11 more). 12 tools only read.
1 tool reads private data (search_context); 4 tools bring in text other people can write (accept_invite, invite, search_context and 1 more); 3 tools can send data out of your system (accept_invite, decline_invite and invite). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.
The fixed checks matched 1 finding (tool-shadowing). Each one comes with the text it matched.
Its remote endpoint lists its tools without sign-in, and its tools changed 2 times since Anlyon Radar first saw them.
Last scanned Nov 10, 2026, version 1.1.2.
Tools
Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.
33 tools
Deletes or cancels 3Claude Code: deny
delete_factSoft-delete a stored fact by its UUID (Smart Forget).No findings
Why it is in this group
- The name "delete_fact" says "delete": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Soft-delete a stored fact by its UUID (Smart Forget). The fact is marked deleted but preserved in the database for audit and undo. Excluded from all normal searches. Use when information is no longer accurate or relevant.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
leave_grantLeave an active namespace grant as the grantee (stop using shared namespace).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Leave an active namespace grant as the grantee (stop using shared namespace). Server verifies grantee authorization. Grantors who want to cut off a collaborator must use revoke_grant.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
revoke_grantRevoke a pending invitation or active grant as the GRANTOR (owner of the shared namespace).No findings
Why it is in this group
- The name "revoke_grant" says "revoke": a word for removing or ending something.
- The server marks it destructive (destructiveHint: true).
Description, as the server wrote it Revoke a pending invitation or active grant as the GRANTOR (owner of the shared namespace). Idempotent. Grantees use decline_invite / leave_grant.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: true
- idempotentHint: true
- openWorldHint: false
Can't be undone 3Claude Code: ask
accept_inviteAccept a namespace invitation addressed to THIS authenticated account's verified email (two-step email OTP).Reads text others writeCan send data outNo findings
Why it is in this group
- The name "accept_invite" says "invite": a word for sending, and what is sent cannot be taken back.
- The input takes a statement or command to run (input.code), which can change anything it can reach.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe name "accept_invite" says "invite": it sends or publishes something to other people.
Description, as the server wrote it Accept a namespace invitation addressed to THIS authenticated account's verified email (two-step email OTP). Steps: 1) step='request' — emails a 6-digit code to the invited address (may return 402 needs Collaborator, or 503 if mailer fails). 2) step='confirm' with code= the 6 digits — activates the grant. Then switch_workspace(name=<grantor>) enters it on this same connection (mcp_config with the X-Ctxstore-Workspace header is also returned for header-based clients). The code activates the named grant_id only; it authorizes nothing else. Obtain grant_id from list_invites(). Docs: https://ctxstore.ai/docs.html#verify-invitations
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
decline_inviteDecline a pending (status=invited) namespace invitation as the grantee.Can send data outNo findings
Why it is in this group
- The name "decline_invite" says "invite": a word for sending, and what is sent cannot be taken back.
What it can reach
- outboundThe name "decline_invite" says "invite": it sends or publishes something to other people.
Description, as the server wrote it Decline a pending (status=invited) namespace invitation as the grantee. Server verifies you are the invitee (email/tenant match). Grantors must use revoke_grant instead.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
inviteInvite one person, by email, to read and write one namespace prefix of THIS account (for example project:barn).Reads text others writeCan send data outNo findings
Why it is in this group
- The name "invite" says "invite": a word for sending, and what is sent cannot be taken back.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
- outboundThe name "invite" says "invite": it sends or publishes something to other people.
Description, as the server wrote it Invite one person, by email, to read and write one namespace prefix of THIS account (for example project:barn). Creates a pending grant and emails them; nothing is shared until they accept with list_invites and accept_invite from their own account. Calling again for the same person and prefix returns the existing invitation without a second email. Withdraw with revoke_grant(grant_id). Not available from a shared workspace session. Their first 30 days of Collaborator time are free; after that you can pay their $19 per 30 days from /account. You are emailed when they accept, and list_workspaces shows whether they have entered yet. Docs: https://ctxstore.ai/docs.html#verify-invitations
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: true
Changes things 15Claude Code: ask
bind_agentBind a sticky agent identity to this MCP session.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Bind a sticky agent identity to this MCP session. Once bound, wake_status and load_context default to this agent_id when called without an explicit agent_id= arg — no need to pass identity on every call. Per-call override still works. Persisted to mcp_sessions.agent_id (migration 0008), so the binding survives setup_account(mode='restore') and the server re-hydrates it automatically on the next tool call. Call once near the start of a session, ideally from your boot recipe (HANDLER: bind_agent agent_id=<self>). Idempotent: re-binding the same agent_id is a no-op. Pass agent_id='' to clear the binding.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
ghost_modeEnable or disable ghost mode for this session.No findings
Why it is in this group
- The description's first verb is "enable": a word for a change.
Description, as the server wrote it Enable or disable ghost mode for this session. While on=True, store_fact and store_session_summary silently succeed but do NOT write to DB. search_facts, load_context, and search_context always work normally. Ghost mode resets to off on MCP session reconnect (not persisted).
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
index_statsUsage statistics from the tenant activity index: total operations, breakdown by agent, by namespace, and by day.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Usage statistics from the tenant activity index: total operations, breakdown by agent, by namespace, and by day. Useful for understanding which agents are most active and where.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
introduce_yourselfOptional: introduce yourself to the platform.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Optional: introduce yourself to the platform. Stores your model info and a brief note in the shared agent registry. Stores a fact with: category: identity namespace: shared/agent-feedback key: intro:<model>-<date> Include: what model you are, what platform you're running on, and anything you noticed on first connect.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
load_contextLoad your bootstrap context deterministically — pulls L0 (identity) + L1 (architecture) + latest L2 (state) + latest L3 (session summary).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Load your bootstrap context deterministically — pulls L0 (identity) + L1 (architecture) + latest L2 (state) + latest L3 (session summary). Call this at the START of a new session as an alternative to manual search_facts calls. Returns your layered memory structured for immediate use. Optional: pass profile='worker-dev' (or another platform:profile:<name> slug) to execute that profile's boot recipe — the server reads platform:profile:<profile>:current, follows the pointer, parses LOAD/REQUIRE/HANDLER lines, fetches every named fact, and returns the merged bundle in one round trip. vars= supplies values for $TASK_FACT_KEY and similar recipe variables.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
mark_salientSalience peptide — temporarily boost a fact's retrieval ranking because it matters RIGHT NOW: an approaching deadline, a live incident, the topic under active work.No findings
Why it is in this group
- The name "mark_salient" says "mark": a word for a change.
Description, as the server wrote it Salience peptide — temporarily boost a fact's retrieval ranking because it matters RIGHT NOW: an approaching deadline, a live incident, the topic under active work. The boost is small and FAST-DECAYING (max 1.4x, ~6-hour half-life — gone by tomorrow), unlike reinforce_fact which rewards confirmed outcomes on a weeks-long timescale. Name the current situation in `reason`. Salience is ranking-only: it can never override tenant isolation, namespace grants, or privacy filters, and stacked peptide boosts are globally capped.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
move_factMove a fact to a different namespace or layer without changing its content.No findings
Why it is in this group
- The name "move_fact" says "move": a word for a change.
Description, as the server wrote it Move a fact to a different namespace or layer without changing its content. Updates metadata in place — preserves fact_id, text, embedding, timestamp, and deprecation chain. Use for reclassifying facts (e.g. bootstrap-v2 refactor) without the storage bloat of store-new + soft-delete-old.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
reinforce_factReward peptide — reinforce a fact whose recall LED TO A REAL GOOD OUTCOME (fix worked, deploy green, user approved).No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Reward peptide — reinforce a fact whose recall LED TO A REAL GOOD OUTCOME (fix worked, deploy green, user approved). Boosts its future retrieval ranking with a bounded, time-decaying multiplier (max 1.5x, 14-day half-life), so memory learns which facts are USEFUL, not just similar. Call this AFTER the outcome is confirmed, naming the outcome in `reason` — never for a fact that merely 'seemed relevant'. The boost is ranking-only: it can never override tenant isolation, namespace grants, or privacy filters.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
resume_sessionCall to load session continuity.No findings
Why it is in this group
- The description's first verb is "call": a word for a change.
Description, as the server wrote it Call to load session continuity. Last session: no prior session.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
setup_accountSet up or restore your persistent memory connection.Reads text others writeNo findings
Why it is in this group
- The description's first verb is "set": a word for a change.
- The input takes a statement or command to run (input.code), which can change anything it can reach.
What it can reach
- untrusted-inputThe server marks it openWorldHint: true: it reaches things outside the server's own data.
Description, as the server wrote it Set up or restore your persistent memory connection. Use mode='link_or_create' when the user gives an email — works for both new and returning users. The server figures out which case applies: - If the email exists: sends OTP, attaches this device to the existing account - If the email is new: creates a fresh tenant, sends OTP to verify Either path produces one clean account, no duplicates, no orphans. Call setup_account(mode='link_or_create', email='their@email.com'). Then ask the user for the 6-digit code from their email and call setup_account(mode='verify', code='123456', email='their@email.com'). Passing email= on verify is required — session IDs rotate between requests on some clients. mode='restore': use when the user provides an emk_* API key directly. mode='reveal_key': echo the full emk_* API key. Works directly on an ALREADY-authenticated session; if the session isn't bound (some clients rotate or omit session ids between requests, so this can happen right after a successful verify), pass email= to receive a 6-digit code, then call again with email= and code= — the same rotation-proof lookup as mode='verify'. Success responses only show a key fingerprint (first 8 chars + last 4) — call this mode only when the user actually needs the full key, e.g. to persist it as 'Authorization: Bearer <key>' in an MCP connector config. Callers who prove neither a bound session nor email ownership are refused. If no email yet, ask: 'What email should I use for your memory account?' After setup completes, call search_facts as the first action to load prior context.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: true
store_factStore information that should persist across sessions.No findings
Why it is in this group
- The name "store_fact" says "store": a word for a change.
Description, as the server wrote it Store information that should persist across sessions. Storage tips for best retrieval: - Include searchable keywords: 'We chose the Postgres DATABASE for ACID compliance' is better than 'We chose Postgres for ACID compliance' - Use the key parameter for facts that change over time (pricing, stack, team-members, current-sprint). When you store with the same key, the old version is archived automatically. - Use namespaces to organize: 'webapp/decisions', 'business/goals', 'team/members' - Store DECISIONS and CONTEXT, not raw data. 'We decided X because Y' is more useful than 'X happened' - Use category to classify: technical, decision, preference, identity, relationship - Use layer for context depth: 0=identity(stable), 1=architecture(monthly), 2=state(weekly), 3=session(daily) Before ending a conversation, call store_session_summary to preserve continuity.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
store_factsStore up to 50 facts in one call — the same fields as store_fact per item.1 finding
Why it is in this group
- The name "store_facts" says "store": a word for a change.
Findings
- Warn
The description contains "every other tool", wording about other tools or servers.
What this check looks forMatched text
description: "...en known, so the memory follows them to every other tool. Each item is stored independently; the..."
Description, as the server wrote it Store up to 50 facts in one call — the same fields as store_fact per item. Made for carrying memory over: when you already remember things about the person from your own platform memory (ChatGPT memory, Claude memory, Gemini saved info), write them here one fact per item, in their words, with the date learned when known, so the memory follows them to every other tool. Each item is stored independently; the reply lists what landed and what did not.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
store_session_summaryStore a compressed summary of this session to preserve continuity for the next one.No findings
Why it is in this group
- The name "store_session_summary" says "store": a word for a change.
Description, as the server wrote it Store a compressed summary of this session to preserve continuity for the next one. Call this before ending a conversation — it is the dream cycle. Stores a L3 session fact with key='session-latest' (or 'session-latest:claude-code' etc.), auto-archiving the previous version. The next session loads this via load_context(). Include: what was discussed/decided, what changed, what's next/unresolved. Collaborative takeaways: at natural session-end signals (user says 'okay I'm out', 'let's wrap', 'bye'), draft a one-line takeaway and confirm with the user before storing. Example: "I'll remember: we fixed the rebind bug and decided to skip transcript ingestion for MVP. Sound right?" Edit based on their response, then call store_session_summary. This makes memory consensual and catches mistakes immediately. Use session_id to scope by client: 'claude-code', 'claude-chat', 'cursor', 'openclaw'
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: false
- openWorldHint: false
switch_workspaceNarrow this session's memory to one project: name='<slug>' scopes recall, the wake bundle, self-notes and the default namespace for new facts to project:<slug>:* (identity and conv…No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
Description, as the server wrote it Narrow this session's memory to one project: name='<slug>' scopes recall, the wake bundle, self-notes and the default namespace for new facts to project:<slug>:* (identity and conventions always ride along). Returns the narrowed bundle right away. name='' widens back to the whole account. Sticky for the session and across reconnects; bind_agent is unaffected. The same call ENTERS a workspace someone shared with this account: name='<their name>' (or their tenant id, or the granted prefix, as listed by list_workspaces) switches this connection into it; no second server, no header, no restart; it holds for the session, and a new session re-enters with the same call. name='' leaves it. When a person asks to work in someone else's shared memory, call this.
Annotations the server declares
- readOnlyHint: false
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
watch_eventsLong-poll this account's fact-write events instead of re-scanning with since_epoch.No findings
Why it is in this group
- Nothing in its name or description says what it does, so it is counted as a write.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Description, as the server wrote it Long-poll this account's fact-write events instead of re-scanning with since_epoch. Every store/supersede/delete/archive/restore/move publishes one event {key, id, namespace, layer, op, ts, seq}. One call waits up to max_wait_s (max 25) for the first matching event and returns up to 50, plus next_since. MCP has no streaming here, so loop: pass next_since back as `since` on the next call and nothing is missed (the server keeps the last ~10000 events per account). Omit since to start from now; since=0 replays what is retained; a bare integer N returns events with seq > N. EXAMPLES - Wait for hands to post results: watch_events(prefix='task:pr27-review:result:', max_wait_s=25) - Resume: watch_events(prefix='comms:<your-agent-id>:', since='<next_since>') Inside a workspace only events for keys the grant can read are returned. Long-lived clients can use GET /api/v1/events (SSE) instead.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
Only reads 12Show 12 read-only tools
get_factRetrieve a fact by its exact key — deterministic, never misses.No findings
Why it is in this group
- The name "get_fact" says "get": a word for reading.
Description, as the server wrote it Retrieve a fact by its exact key — deterministic, never misses. Use get_fact for self-notes: get_fact(key='agent:<name>:self-note:opening') or get_fact(key='agent:<name>:self-note:closing'). search_facts misses self-notes because their embedding similarity scores below other identity facts — get_fact(key=) is the only reliable retrieval path. Also use for inter-agent handoffs when the exact key is known. Returns the current (non-deprecated) version of the fact.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_fact_historyReturn the full version history for a keyed fact — current version plus all deprecated predecessors.No findings
Why it is in this group
- The name "get_fact_history" says "get": a word for reading.
Description, as the server wrote it Return the full version history for a keyed fact — current version plus all deprecated predecessors. Use to see how a fact evolved over time or to undo an accidental correction. Returns list of fact versions ordered newest → oldest.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_session_seedReturn the pre-computed session seed for this tenant — a compact first-person bio (~500 tokens) describing who you are and what you care about right now.No findings
Why it is in this group
- The name "get_session_seed" says "get": a word for reading.
Description, as the server wrote it Return the pre-computed session seed for this tenant — a compact first-person bio (~500 tokens) describing who you are and what you care about right now. Generated nightly and on session disconnect. Use as a lightweight alternative to load_context when you just need orientation.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
get_statsGet your memory usage statistics — total facts stored, plan limits, collection sizes, and daily query counts.No findings
Why it is in this group
- The name "get_stats" says "get": a word for reading.
Description, as the server wrote it Get your memory usage statistics — total facts stored, plan limits, collection sizes, and daily query counts. Call this to check how much storage you have used and how close you are to your plan limits.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
index_recentReturn the most recent fact operations across all agents in this tenant, newest first.No findings
Why it is in this group
- The description's first verb is "return": a word for reading.
Description, as the server wrote it Return the most recent fact operations across all agents in this tenant, newest first. Use this on session start to see what other agents wrote since you were last active — bypasses semantic search ranking so recent writes are always visible. Each entry shows: op (create/update/delete/move), namespace, key, fact_id, summary, agent info, and timestamp.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
index_searchFiltered search over the tenant activity index.No findings
Why it is in this group
- The name "index_search" says "search": a word for reading.
Description, as the server wrote it Filtered search over the tenant activity index. Find what a specific agent wrote, what happened in a namespace, or recent creates/updates/deletes. Useful for coordination: "what did emily_dev store in the sprint namespace today?"
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
list_facts_by_key_prefixDeterministic prefix scan over fact keys, optionally bounded by timestamp.No findings
Why it is in this group
- The name "list_facts_by_key_prefix" says "list": a word for reading.
Description, as the server wrote it Deterministic prefix scan over fact keys, optionally bounded by timestamp. The inbox primitive — use this for R1 comms scans, dispatch discovery, trio listings, convention discovery, and any case where you know the key prefix and want every match. Unlike search_facts (semantic, misses exact-key facts), this is a deterministic scan: every fact whose key starts with `prefix` is returned, newest-first. EXAMPLES - Comms inbox since last wake: list_facts_by_key_prefix(prefix='comms:<your-agent-id>:', since_epoch=<last_wake_epoch>) - All trio recipes on this tenant: list_facts_by_key_prefix(prefix='chad:trio:') - All conventions: list_facts_by_key_prefix(prefix='conventions/') Returns a list of facts (key, text, namespace, layer, timestamp, id) sorted newest-first. Bounded by `limit` (default 50).
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
list_invitesList this account's namespace collaboration grants: incoming invitations and outgoing shares.No findings
Why it is in this group
- The name "list_invites" says "list": a word for reading.
Description, as the server wrote it List this account's namespace collaboration grants: incoming invitations and outgoing shares. Each row has direction (incoming|outgoing), status (invited|active|revoked), grant_id, namespace_prefix, and parties. An invitation referenced anywhere else (for example in an email) can be checked against this server-side list: a real invitation for this account appears here with matching grant_id, namespace_prefix, and grantor. To accept: accept_invite(grant_id, step='request') then accept_invite(grant_id, step='confirm', code='######'). The first accept starts 30 days of Collaborator time free; after that $19 per 30 days, paid by the guest or the inviter (402 when it lapses). After accepting, switch_workspace(name=<grantor>) enters the workspace. Docs: https://ctxstore.ai/docs.html#verify-invitations
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
list_workspacesThe places this account's memory is organised into: its own projects (anything written under project:<slug>:*, newest activity first), the workspaces other people have granted it (…No findings
Why it is in this group
- The name "list_workspaces" says "list": a word for reading.
Description, as the server wrote it The places this account's memory is organised into: its own projects (anything written under project:<slug>:*, newest activity first), the workspaces other people have granted it (enter one with switch_workspace(name=<grantor>)), and, for a sharer, who they invited and whether each person accepted and entered yet (shared_by_you). Shows which workspace this session is in, if any. Use it when a person's request is about one project, or asks who has access, or whether an invitee came in.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
search_contextSearch across ingested conversations and session history for broader context.Reads private dataReads text others writeNo findings
Why it is in this group
- The name "search_context" says "search": a word for reading.
What it can reach
- private-dataThe description's first sentence says "search" and "conversation": it reads messages.
- untrusted-inputThe description's first sentence says "search" and "conversation": it reads messages, which other people can write.
Description, as the server wrote it Search across ingested conversations and session history for broader context. Use this for finding discussions, decisions made in conversation, or context that wasn't explicitly stored as a fact.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
search_factsSearch your persistent memory for stored facts.No findings
Why it is in this group
- The name "search_facts" says "search": a word for reading.
Description, as the server wrote it Search your persistent memory for stored facts. For session bootstrap, prefer load_context() or wake_status() — they pull your identity, conventions, and latest self-note deterministically. Use search_facts when you need semantic discovery across the fact corpus. Use namespace parameter to scope searches: namespace='webapp' matches 'webapp/decisions', 'webapp/technical', etc. (prefix matching). Search tips: use the words the fact was stored with. If you stored 'Postgres database', search for 'database' or 'Postgres', not 'data storage solution'.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
wake_statusCheck which context layers loaded successfully on this session start.No findings
Why it is in this group
- The name "wake_status" says "status": a word for reading.
Description, as the server wrote it Check which context layers loaded successfully on this session start. Returns a structured status: which of embryo, conventions, agent identity, closing note, and opening note were found, plus a missing[] list. Call this after load_context() or resume_session() to verify your bootstrap was complete and diagnose what to fetch manually. Optional: pass profile='worker-dev' (or another platform:profile:<name> slug) to additionally check the REQUIRE keys named in that profile's recipe. Missing required facts surface in the same missing[] list.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.
Exfil surface
Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.
12 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 1 private-data tool breaks every one.
- Reads private data1
- Reads files, mail, messages, databases, secrets, calendars or repositories.
- Reads text others write4
- Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
- Can send data out3
- Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
{
"permissions": {
"deny": [
"mcp__ctxstore__search_context"
]
}
}Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.
Remote posture
What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.
Looked at Nov 20, 2026, 18:25 UTC, unchanged since Oct 29, 2026. No credentials were sent and no tool was called. What is looked at.
- Reachable
- yes
- HTTP status of an anonymous initialize
- 200
- tools/list served without credentials
- yes
- Tools that change things, listed without credentials
- 21
- Protocol version
- 2025-06-18
- WWW-Authenticate on 401
- not observed
- resource_metadata in WWW-Authenticate
- not observed
- RFC 9728 protected resource metadata
- https://mcp.ctxstore.ai/.well-known/oauth-protected-resource/mcp
- Authorization servers
- https://ctxstore.ai
- Legacy HTTP+SSE transport
- no
- Dynamic client registration
- yes
- Client ID metadata documents
- no
- HighThe whole server
The endpoint served tools/list without credentials, and the list has 21 tools that change things (4 destructive, 2 irreversible, 15 write). Whether a call also works without credentials was not tested: tools are never called.
What this check looks forMatched text
https://mcp.ctxstore.ai/mcp: tools/list returned 33 tools to an anonymous client, including decline_invite (destructive), delete_fact (destructive), leave_grant (destructive), revoke_grant (destructive), accept_invite (irreversible) and 16 more
Posture changes
- Oct 29, 2026
- Tools that change things, listed without credentials: 20 to 21
Findings
Matches of fixed checks, with the text they matched.
- Warnstore_facts
The description contains "every other tool", wording about other tools or servers.
What this check looks forMatched text
description: "...en known, so the memory follows them to every other tool. Each item is stored independently; the..."
History
Changes to its tool definitions and to its remote endpoint, newest first.
Tools5 changed.~5 changedView diff
- Changed:get_factOnly readsinputSchema
- Changed:inviteCan't be undoneinputSchema
- Changed:search_factsOnly readsinputSchema
- Changed:store_factChanges thingsinputSchema
- Changed:store_factsChanges thingsinputSchema
PostureRemote posture changed. Tools that change things, listed without credentials: 20 -> 21.View diff
- Tools that change things, listed without credentials: 20 to 21
Tools1 added, 4 changed.+1 added~4 changedView diff
- Added:inviteCan't be undone
- Changed:accept_inviteCan't be undonedescription
- Changed:list_invitesOnly readsdescription
- Changed:list_workspacesOnly readsdescription
- Changed:switch_workspaceChanges thingsdescriptioninputSchema
Install
Pinned to version 1.1.2. Values in angle brackets are yours to fill in.
claude mcp add --transport http ctxstore https://mcp.ctxstore.ai/mcpclaude mcp add --transport http ctxstore https://mcp.ctxstore.ai/mcpRun in a terminal.
Claude Code permissions
Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.
{
"permissions": {
"allow": [
"mcp__ctxstore__get_fact",
"mcp__ctxstore__get_fact_history",
"mcp__ctxstore__get_session_seed",
"mcp__ctxstore__get_stats",
"mcp__ctxstore__index_recent",
"mcp__ctxstore__index_search",
"mcp__ctxstore__list_facts_by_key_prefix",
"mcp__ctxstore__list_invites",
"mcp__ctxstore__list_workspaces",
"mcp__ctxstore__search_context",
"mcp__ctxstore__search_facts",
"mcp__ctxstore__wake_status"
],
"ask": [
"mcp__ctxstore__accept_invite",
"mcp__ctxstore__bind_agent",
"mcp__ctxstore__decline_invite",
"mcp__ctxstore__ghost_mode",
"mcp__ctxstore__index_stats",
"mcp__ctxstore__introduce_yourself",
"mcp__ctxstore__invite",
"mcp__ctxstore__load_context",
"mcp__ctxstore__mark_salient",
"mcp__ctxstore__move_fact",
"mcp__ctxstore__reinforce_fact",
"mcp__ctxstore__resume_session",
"mcp__ctxstore__setup_account",
"mcp__ctxstore__store_fact",
"mcp__ctxstore__store_facts",
"mcp__ctxstore__store_session_summary",
"mcp__ctxstore__switch_workspace",
"mcp__ctxstore__watch_events"
],
"deny": [
"mcp__ctxstore__delete_fact",
"mcp__ctxstore__leave_grant",
"mcp__ctxstore__revoke_grant"
]
}
}Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as ctxstore. Review it before you use it.
Capability tags behind 5 rules
- mcp__ctxstore__accept_invite untrusted-input, outbound
- mcp__ctxstore__decline_invite outbound
- mcp__ctxstore__invite untrusted-input, outbound
- mcp__ctxstore__search_context private-data, untrusted-input
- mcp__ctxstore__setup_account untrusted-input
Pin
Keep the decision: fail CI when the tools change, follow every change, or show a badge.
Pin in CI
Fail a build when the tools you reviewed are not the tools the server serves today.
curl -s "https://radar.anlyon.com/api/v1/servers/ai.ctxstore/ctxstore-mcp/verify?hash=58b0f0cfea0e3b9dc1e32eedcedc5a9a4c4f2985bbe8e778683809592e53cec2" | jq -e '.changed == false'Pinned hash 58b0f0cfea0e, from Nov 10, 2026, 01:18 UTC.
Watch
Get told when its tools change: a new tool, a new description, a new class.
https://radar.anlyon.com/servers/ai.ctxstore/ctxstore-mcp/changes.xmlAn Atom feed. Works in any feed reader or Slack's RSS app.
Badge
[](https://radar.anlyon.com/servers/ai.ctxstore/ctxstore-mcp)Source: MCP Registry · repository. Synced Oct 21, 2026, 17:41 UTC.
ctxstore — memory that follows you across models MCP server: questions and answers
Answered from the scan above, not from the publisher's description.
Is the ctxstore — memory that follows you across models MCP server safe?
- No scan can promise that. What its definitions show: 33 tools, 21 of which can change, send or delete something, and 1 finding from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.
Can the ctxstore — memory that follows you across models MCP server delete data?
- Yes. 3 tools can delete, cancel or revoke something: delete_fact, leave_grant and revoke_grant.
Can the ctxstore — memory that follows you across models MCP server do things that cannot be undone?
- Yes. 3 tools can send, pay, publish or merge: accept_invite, decline_invite and invite.
What tools does the ctxstore — memory that follows you across models MCP server have?
- 33 tools: 3 destructive, 3 irreversible, 15 write, 12 read. delete_fact, leave_grant, revoke_grant, accept_invite, decline_invite, invite, bind_agent, ghost_mode and 25 more.
Does the ctxstore — memory that follows you across models MCP server need sign-in or OAuth?
- Its remote endpoint listed its tools to an anonymous client, without sign-in.
Does the ctxstore — memory that follows you across models MCP server send data to outside services?
- 3 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): accept_invite, decline_invite and invite.
Has the ctxstore — memory that follows you across models MCP server changed its tools?
- Yes, 2 times since Anlyon Radar started tracking it, most recently on Nov 10, 2026. Every change is on this page and in its Atom feed.