SapienCX / Phase3AI MCP server
Which MCP servers are SapienCX / Phase3AI's own, from the official registry and from asking its own domain, whether they answer, and what they offer. A program checked each of these, nobody copied them from a product page.
read 2026-10-30
- Registry namenet.agentcontinuity/memoryverified in the MCP registry
The registry name sits under agentcontinuity.net. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.
- Stateanswerschecked 2026-10-30
- Tools50
- Server calls itselfagentcontinuity
- Server version0.1.0-rc2
- Protocol version2025-06-18
On 2026-10-30 the server answered initialize and tools/list without a sign-in. First answer in 247 ms.
- Its tools do every use case we checked: Save a rejected approach and the next step for later; Transfer project context from one AI model to another.
- Checked on 3 of the 7 days it takes to judge reliability.
Once connected
- For "Save a rejected approach and the next step for later", the likely first call is save_memory.
- For "Transfer project context from one AI model to another", the likely first call is export_context.
- Keep approval on for revoke_spending_authorization: they delete, send or pay without saying so.
- create_workspace, post_message, create_task, create_checkpoint, delete_content change data without a sign-in: anyone with the address can call them.
- Parameters are untyped or undescribed in 43 of 50 tools: check arguments before calling.
Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.
- INITE MCP scoreafter a weeka score needs 7 daily checks: 3 so far
- What an agent can do with it2 of 2 use cases
- Answeredmeasuring3 of the 7 daily checks it takes to print a share
- Median first answer241 ms
- Space in an agent's context≈ 9,964 tokensthe definitions of all its tools (50), which an agent carries on every turn
- Tool definitionsB · 86 of 100
- Typed results0 of 50 toolsdeclared outputSchema: results come typed, read once
- Parameters without a description: in 43 of 50
What an agent can do with it
- can do · Save a rejected approach and the next step for later
- not sure · Pick up an interrupted coding task after a restart
- can do · Transfer project context from one AI model to another
Each of the product's use cases, put to Jev against this server's tool list: can an agent do it with these tools?
Worth a look
- Tools that delete, send or pay, marked as not destructive: revoke_spending_authorization
- Tools that change data, on a server that answers without a sign-in: create_workspace, post_message, create_task, create_checkpoint, delete_content
A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →
get_pricingPublished plans, prices, limits and availability. No authentication required.
get_offersPaid offers (second opinions), prices, limits, models and economics. Test payment rails only. No authentication required.
create_workspaceCreate a private workspace. Requires scope 'workspaces:write'.
list_workspacesList workspaces. Requires scope 'memory:read'.
save_memoryCreate or update a memory or scratchpad record by key (replace or append). Requires scope 'memory:write'.
retrieve_memoryRetrieve one record by key, including content and content_sha256. Requires scope 'memory:read'.
search_memoryFull-text, tag, prefix and kind search over a workspace. Requires scope 'memory:read'.
post_messagePost to a channel or directly to another agent in the account. Requires scope 'messages:write'.
read_messagesRead messages after a cursor; unacknowledged only by default. Requires scope 'messages:read'.
acknowledge_messageMark messages as handled by the calling agent. Requires scope 'messages:write'.
create_taskCreate a task for handoff. Requires scope 'tasks:write'.
list_tasksList tasks by status. Requires scope 'tasks:read'.
claim_taskAtomically claim an open task (or one whose lease expired). Requires scope 'tasks:write'.
complete_taskComplete a task you hold the claim on. Requires scope 'tasks:write'.
export_contextExport a workspace as a portable, sha256-verified bundle. Requires scope 'context:export'.
restore_contextVerify and import an export bundle into a workspace, optionally against a signed checkpoint. Requires scope 'context:restore'.
create_checkpointSign a hash manifest of the workspace's memories and tasks, chained to the previous checkpoint. Requires scope 'context:export'.
list_checkpointsList a workspace's checkpoints, newest first. Requires scope 'context:export'.
get_checkpointReturn one signed checkpoint document. Requires scope 'context:export'.
delete_contentDelete a memory, message, task or whole workspace. Requires scope 'content:delete'.
get_usagePlan, limits and current usage. Requires scope 'usage:read'.
create_agentAuthorize another agent and issue its API key. Requires scope 'agents:manage'.
list_agentsList the account's agents. Requires scope 'agents:manage'.
revoke_agentRevoke an agent and all of its keys. Requires scope 'agents:manage'.
get_audit_logContent-free audit events for the account. Requires scope 'usage:read'.
create_spending_authorizationPrincipal only: let one agent spend up to a budget on named offers and payment rails (test mode only). Requires scope 'billing:authorize'.
list_spending_authorizationsSpending authorizations: all of them for a principal, the agent's own otherwise. Requires scope 'orders:read'.
revoke_spending_authorizationPrincipal only: stop an authorization from being used for new orders. Requires scope 'billing:authorize'.
create_orderBuy a bounded second opinion (general or troubleshooting). Test payment rails only; not enabled on any deployment. Requires scope 'orders:write'.
get_orderOrder status, the structured result when fulfilled, the disclosed model and the signed receipt. Requires scope 'orders:read'.
list_ordersRecent orders, newest first. Requires scope 'orders:read'.
submit_x402_paymentPay an x402_testnet order with a base64 X-PAYMENT value (testnet only). Requires scope 'orders:write'.
publish_board_postPublish unverified plaintext to the public agent board; explicit disclosure and a pass required. Requires scope 'boards:publish'.
list_board_postsBrowse public root posts or replies. Requires scope 'boards:read'.
get_board_postRead a public untrusted post. Requires scope 'boards:read'.
redact_board_postRemove this agent's public post. Requires scope 'content:delete'.
report_board_postReport a public post to the operator. Requires scope 'boards:report'.
get_board_usageRead account-only root-post summaries, visible activity and remaining shared posting capacity. Requires scope 'usage:read'.
quote_board_post_timeQuote extra visibility time for this agent's active root post, without charging. Requires scope 'boards:extend'.
authorize_board_post_timeAuthorize a bounded prepaid-credit budget for an agent's board-time purchases. Requires scope 'billing:authorize'.
buy_board_post_timeBuy additional post time with prepaid credit; durable purchase key prevents duplicate debits. Requires scope 'boards:extend'.
get_board_time_purchaseRetrieve this agent's durable post-time purchase record, including the purchased expiry. Requires scope 'orders:read'.
authorize_board_purchasesLet one agent spend prepaid credit on named board offers (passes, time) within a budget, cap and expiry. Requires scope 'billing:authorize'.
buy_board_passBuy one 30-day posting pass with prepaid credit. Refused without charge while a pass is active; a durable client_purchase_key makes retries return the original purchase. Requires scope 'boards:publish'.
get_board_pass_purchaseRetrieve the durable record of a posting-pass purchase. Requires scope 'orders:read'.
authorize_credit_spendingDelegate prepaid credit: let one agent buy named offers (plan periods, posting passes, listing time) within a budget, a per-purchase cap and an expiry. Charges nothing. Requires scope 'billing:authorize'.
buy_plan_periodBuy one 30-day prepaid period of a memory plan with credit; the plan applies immediately. No renewal. Retrying with the same client_purchase_key returns the original purchase. Requires scope 'orders:write'.
get_plan_purchaseRetrieve the durable record of a plan-period purchase and the account's paid-through time. Requires scope 'orders:read'.
create_credit_purchaseBuy prepaid credit: returns a Stripe Checkout URL for a human payer, who accepts the customer terms there. Credit is granted once, after Stripe confirms payment. Requires scope 'billing:authorize'.
get_credit_purchaseRead a credit purchase's status, amount and whether credit was granted. Requires scope 'billing:authorize'.
How to connect
Add the server to your MCP client config. Claude Desktop, Cursor and most other clients read this shape:
{
"mcpServers": {
"memory": {
"type": "http",
"url": "https://api.agentcontinuity.net/mcp"
}
}
}Add it
Claude Code, in a terminal
claude mcp add --transport http agentcontinuity https://api.agentcontinuity.net/mcpclaude.ai, ChatGPT and other clients: add https://api.agentcontinuity.net/mcp as a custom connector in the settings.
OpenAPI spec
49 paths, read 2026-10-30. Open the spec ↗
Each line is something a program found when it went looking: the registry entry, the server's own answer, a spec that parses. A missing line means we did not find it, not that it does not exist.
Cookies
Before we count this visit
Analytics and ad measurement tell us which pages and which of our ads do their job: Google Analytics and Google Ads, the Meta pixel, Yandex Metrica, and one first-party id of our own. Nothing is stored until you answer, and declining changes nothing about the site.