// MCP TOKEN COST

The Stable Baseline MCP server costs 100,788 tokens before the first call.

Every request your agent makes carries every tool definition this server exposes — context your code, documents and conversation can't use, mostly for tools the agent never calls. You don't need them all in the window, and you don't have to pay for them.

QUICK ANSWER The Stable Baseline MCP server's 222 tool definitions consume 100,788 tokens — 50% of a 200k context window, and 54× the median MCP server (1,866 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS tiktoken o200k_base · rank #19 of 12,802 measured servers · refreshed every build Method →

// CONTEXT WINDOW SHARE

What that costs before your agent starts working.

Tool definitions are overhead: they occupy context on every request and compete with your code, documents and conversation history for the same window.

200K WINDOW 50%

1M WINDOW 10%

Corpus context: Stable Baseline ranks #19 of 12,802 measured MCP servers by definition cost. The median is 1,866 tokens, p90 is 13,667, and the heaviest (Ainumbers Mcp Apps) is 445,124 — 223% of a 200k window on its own. New to this? See MCP token cost and context window in the glossary.

Token cost is one axis. See the risk picture across your whole stack →

// PER-TOOL BREAKDOWN

Where the 100,788 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 454 tokens per tool.

Tool	Category	Tokens	% of server
updateImprovement	Write	5,910	5.9%
addWhiteboardElements	Write	4,852	4.8%
updateTask	Write	4,494	4.5%
createImprovement	Write	3,332	3.3%
updateDatabase	Write	2,891	2.9%
queryDatabaseRows	Read	2,800	2.8%
autoDesignWhiteboard	Destructive	2,553	2.5%
createTask	Write	2,242	2.2%
designDeckInWhiteboard	Execute	2,100	2.1%
insertDiagramInDocument	Write	2,080	2.1%
kg_search	Read	1,921	1.9%
updateDiagramInDocument	Write	1,826	1.8%
renderDiagram	Execute	1,706	1.7%
saveDatabaseDashboard	Write	1,666	1.7%
updatePlanPhase	Write	1,424	1.4%
saveDatabaseForm	Write	1,410	1.4%
writeDatabaseRows	Write	1,376	1.4%
listImprovements	Read	1,335	1.3%
createDatabase	Write	1,308	1.3%
designIllustrationInWhiteboard	Execute	1,216	1.2%
delegateToAgent	Destructive	1,201	1.2%
editDocument	Write	1,155	1.1%
listDocuments	Read	1,064	1.1%
updateWhiteboardScene	Write	972	1.0%
insertWhiteboardDiagram	Write	967	1.0%
setUpAgentDuty	Write	965	1.0%
previewKgScopeChange	Destructive	925	0.9%
askAgent	Read	921	0.9%
setUpAgent	Write	882	0.9%
listWhiteboardStencils	Read	792	0.8%
listPlans	Read	757	0.8%
insertWhiteboardImage	Write	710	0.7%
startMeetingScribe	Execute	697	0.7%
createPlanPhase	Write	681	0.7%
designComponent	Write	661	0.7%
reorderDocuments	Destructive	653	0.6%
listAssignablePrincipals	Read	623	0.6%
listDocumentVersions	Read	607	0.6%
exportFromWhiteboard	Write	606	0.6%
listTasks	Read	585	0.6%
createBrandKit	Write	573	0.6%
getDatabase	Read	569	0.6%
getWhiteboardImage	Read	549	0.5%
listWhiteboards	Read	548	0.5%
getDocument	Read	544	0.5%
updatePlan	Write	534	0.5%
insertImageInDocument	Write	533	0.5%
importDatabaseRows	Write	529	0.5%
getDeckReplyInWhiteboard	Read	524	0.5%
updateUserPreferences	Write	512	0.5%
updateOrgFeatureFlags	Write	507	0.5%
getAgentRun	Destructive	503	0.5%
deleteDatabaseRows	Destructive	489	0.5%
traceImage	Read	487	0.5%
listAgents	Read	483	0.5%
getAgentConversation	Read	478	0.5%
duplicateWhiteboardElements	Write	455	0.5%
createPlan	Write	424	0.4%
getDashboardImage	Read	421	0.4%
listDatabases	Read	416	0.4%
dataToTable	Read	415	0.4%
getImprovement	Read	408	0.4%
exportDatabaseRows	Write	404	0.4%
exportDocument	Write	398	0.4%
getTask	Read	391	0.4%
findAndReplaceTextInDocument	Write	376	0.4%
approveAgentAction	Write	362	0.4%
upsertResourcePermission	Write	361	0.4%
getDatabaseRow	Read	354	0.4%
listArchitectureIcons	Read	352	0.3%
updateFolder	Write	348	0.3%
searchImprovements	Read	341	0.3%
listDiagramTypes	Read	332	0.3%
createDocument	Write	331	0.3%
editWhiteboardImageRegion	Write	330	0.3%
getDiagramImage	Read	328	0.3%
addPlanActivity	Write	307	0.3%
listResourcePermissions	Read	294	0.3%
kg_get_entity	Read	292	0.3%
getWhiteboard	Read	290	0.3%
listTeams	Read	273	0.3%
createImageUploadSession	Write	273	0.3%
startSignup	Execute	265	0.3%
createDocumentIngestSession	Write	250	0.2%
getEffectivePermission	Read	249	0.2%
getFolderHierarchy	Read	246	0.2%
previewSubscriptionChange	Destructive	242	0.2%
listWorkspaces	Read	234	0.2%
createTaskDependency	Write	231	0.2%
rebuildPlatformCatalogEmbeddings	Write	231	0.2%
inviteMember	Write	230	0.2%
getDatabaseGuide	Read	228	0.2%
updateImageInDocument	Write	226	0.2%
getKgScopeTree	Read	223	0.2%
searchInfographicTemplates	Read	222	0.2%
createDocumentFromUpload	Write	221	0.2%
applyTaskDependencyCascade	Write	220	0.2%
getPlanPhase	Read	219	0.2%
createWhiteboard	Write	216	0.2%
deleteDatabase	Destructive	215	0.2%
addImprovementEvidence	Write	215	0.2%
getDiagramTypeGuide	Read	214	0.2%
getPlan	Read	212	0.2%
addImprovementActivity	Write	210	0.2%
updateProfile	Write	199	0.2%
getProjectHierarchy	Read	198	0.2%
setResourcePermissionOverride	Write	197	0.2%
getDashboardVisualGuide	Read	196	0.2%
kg_evaluate_retrieval	Write	194	0.2%
setPlanItemParent	Write	190	0.2%
deletePlanPhase	Destructive	189	0.2%
createProject	Write	184	0.2%
listFolders	Read	183	0.2%
setDefaultBrandKit	Write	181	0.2%
pollSignupStatus	Read	178	0.2%
createVegaDataUploadSession	Write	176	0.2%
updateOrgSettings	Write	175	0.2%
getDiagramInDocument	Read	172	0.2%
listProjects	Read	162	0.2%
deleteImageInDocument	Destructive	161	0.2%
getMeetingScribeStatus	Read	160	0.2%
reorderFolders	Write	160	0.2%
listTaskDependencies	Read	156	0.2%
updateProject	Write	156	0.2%
deleteDiagramInDocument	Destructive	148	0.1%
listDashboardVisualTypes	Read	148	0.1%
getPlanHierarchy	Read	146	0.1%
getOrgSettings	Read	145	0.1%
listMembers	Read	145	0.1%
createOrganisation	Write	145	0.1%
getPriceForTier	Read	144	0.1%
listInvitations	Read	144	0.1%
createImprovementCategory	Write	141	0.1%
quoteCreditPackage	Execute	140	0.1%
updateMemberRole	Write	140	0.1%
deleteResourcePermission	Destructive	136	0.1%
updateImprovementComment	Write	134	0.1%
createTeam	Write	133	0.1%
updatePlanComment	Write	132	0.1%
removeMember	Destructive	131	0.1%
updateResourcePermission	Write	130	0.1%
getSubscription	Read	127	0.1%
searchTools	Read	127	0.1%
setKgProjectVisibility	Write	126	0.1%
updateTaskDependency	Write	126	0.1%
deleteVegaDataFile	Destructive	124	0.1%
updateOrganisation	Write	124	0.1%
cancelInvitation	Write	123	0.1%
setKgWorkspaceScope	Write	123	0.1%
deletePlan	Destructive	120	0.1%
setMemberActive	Write	120	0.1%
updateTeam	Write	120	0.1%
resendInvitation	Write	118	0.1%
listCreditPurchases	Read	118	0.1%
createWorkspace	Write	117	0.1%
addWorkspaceMember	Write	116	0.1%
applySubscriptionChange	Financial	116	0.1%
stopMeetingScribe	Execute	115	0.1%
previewSubscriptionCancellation	Financial	113	0.1%
getDocumentIngestJob	Read	112	0.1%
applyKgScopeChange	Write	111	0.1%
previewTaskDependencyCascade	Read	107	0.1%
getTeam	Read	107	0.1%
listInvoices	Read	107	0.1%
deleteImprovement	Destructive	106	0.1%
updateWorkspaceMember	Write	104	0.1%
cancelAllKgInScope	Destructive	102	0.1%
getImageInDocument	Read	102	0.1%
previewKgRebuild	Read	100	0.1%
updateTeamWorkspaceAccess	Write	97	0.1%
getOrganisation	Read	96	0.1%
setKgDocumentScope	Write	96	0.1%
purchaseCreditPackage	Financial	95	0.1%
getWhiteboardGuide	Read	95	0.1%
setKgFolderScope	Write	95	0.1%
getCurrentPlanEntitlements	Read	94	0.1%
getMember	Read	94	0.1%
grantTeamWorkspaceAccess	Write	94	0.1%
getCreditBalance	Read	93	0.1%
updateWorkspace	Write	93	0.1%
reactivateSubscription	Financial	91	0.1%
reorderPlanPhases	Write	91	0.1%
acceptTaskDependencyReview	Write	90	0.1%
deleteTaskDependency	Destructive	90	0.1%
listOrganisations	Read	89	0.1%
listPaymentMethods	Read	88	0.1%
reorderImprovementCategories	Write	88	0.1%
resetDocumentInBrain	Destructive	86	0.1%
getCurrentUser	Read	86	0.1%
updateImprovementCategory	Write	86	0.1%
getCustomerPortalLink	Financial	85	0.1%
cancelSubscription	Financial	83	0.1%
getProject	Read	82	0.1%
kg_list_communities	Read	81	0.1%
addTeamMember	Write	81	0.1%
updateBillingEmail	Write	79	0.1%
kg_related_documents	Read	78	0.1%
deleteWhiteboard	Destructive	76	0.1%
removeWorkspaceMember	Destructive	74	0.1%
getWorkspace	Read	74	0.1%
listBrandKits	Read	74	0.1%
cancelKgBuildBatch	Destructive	73	0.1%
deleteImprovementComment	Destructive	73	0.1%
deletePlanComment	Destructive	71	0.1%
revokeTeamWorkspaceAccess	Destructive	69	0.1%
kg_suggest_sample_questions	Write	69	0.1%
dismissTaskDependencyReview	Write	68	0.1%
removeTeamMember	Destructive	68	0.1%
getUserPreferences	Read	68	0.1%
kg_get_wiki_page	Read	68	0.1%
createFolder	Write	68	0.1%
deleteDocument	Destructive	66	0.1%
triggerKgRebuild	Execute	66	0.1%
getCreditPackages	Read	65	0.1%
deleteTeam	Destructive	64	0.1%
kg_scope_status	Read	63	0.1%
listPlanPhases	Read	56	0.1%
listImprovementCategories	Read	46	0.0%
kg_backlinks	Read	44	0.0%
deleteFolder	Destructive	43	0.0%
deleteImprovementCategory	Destructive	43	0.0%
getCdmdLanguageGuide	Read	40	0.0%

// CUT THE COST

Your agent uses a handful of these tools. It pays for all 222.

You don't need all 222 of those definitions in the window. PolicyLayer is an MCP gateway that sits in front of Stable Baseline: only the tools you grant are exposed to the agent, the rest never load. A smaller window means a sharper agent — less noise when it picks a tool — and every request costs less:

Grant scope	Definition cost	Reduction
All 222 tools (no gateway)	100,788 tokens	—
3 granted tools	~1,362 tokens	−99%
5 granted tools	~2,270 tokens	−98%
10 granted tools	~4,540 tokens	−95%

The risk dividend: 36 of these 222 tools are critical-risk (destructive or financial) and cost 9,477 tokens (9% of the definition load). Block them — the recommended starter policy — and you reclaim that context before tuning anything else.

Create a free account and register Stable Baseline — nothing to install.
Grant only the tools you use — ungranted definitions never enter the context window.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

CUT STABLE BASELINE TOKEN COST →

Instant setup, no code required.

Model your own stack in the token-cost calculator, or see the Stable Baseline policy for what a sensible grant looks like.

//FAQ

Stable Baseline token-cost questions.

How many tokens does the Stable Baseline MCP server use?+

Its 222 tool definitions total 100,788 tokens — 50% of a 200k context window — measured with tiktoken o200k_base over the serialised tools/list payload. Exact counts vary slightly by client and model.

Why does Stable Baseline consume tokens before I send a message?+

MCP clients load every connected server's tool definitions — name, description, and input schema — into the model's context so it knows what it can call. That payload is charged against your context window on every request, whether or not a tool is used.

How do I reduce Stable Baseline's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Stable Baseline to only the tools you allow — ungranted definitions are filtered out of the tool list, so they never enter the context window. A grant of 3 typical tools costs roughly 1,362 tokens, a 99% reduction.

Does deferred tool loading fix this?+

Partially, in some clients. Claude Code defers MCP tool schemas behind a tool-search step by default, and VS Code has experimental grouping — but you still pay tokens per search and reload, and Cursor, Windsurf and Gemini CLI load definitions upfront. Reducing the exposed tool set cuts the cost in every client.

// METHOD & REVIEW

How these numbers were measured.

01

Serialisation

Each tool is serialised as a tools/list entry — name, description, input schema — from the schemas in the PolicyLayer scan database. Clients differ slightly in framing, so treat counts as close estimates.

02

Tokeniser

tiktoken o200k_base (GPT-4o/o-series). Anthropic's current tokeniser isn't published, so Claude's exact counts will differ; for English text and JSON schemas the totals are close enough to treat these as estimates.

03

Deferred loading

Some clients now defer schema loading (Claude Code's tool search; VS Code experimental grouping). You still pay per search and reload — and Cursor, Windsurf and Gemini CLI load everything upfront.

Computed 23-10-2026 from the PolicyLayer scan database over all 222 catalogued Stable Baseline tools. Counts refresh with every site build.