// MCP TOKEN COST

The Courier MCP server costs 16,505 tokens before the first call.

Connect Courier and its 123 tool definitions are loaded into the model's context on every request — 8.3% of a 200k window spent before your agent does anything.

QUICK ANSWER The Courier MCP server's tool definitions consume 16,505 tokens — 8.7× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 123 tools · 16,505 tokens · 8.3% of 200k · 1.7% of 1M Method →

// CONTEXT WINDOW SHARE

What that buys 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 8.3%

1M WINDOW 1.7%

Corpus context: Courier ranks #78 of 3,213 measured MCP servers by definition cost. The median is 1,905 tokens, p90 is 7,952, and the heaviest (Fusionauth) is 183,337 — 92% of a 200k window on its own.

// PER-TOOL BREAKDOWN

Where the 16,505 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 134 tokens per tool.

Tool	Category	Tokens	% of server
invoke_ad_hoc_automation	Execute	789	4.8%
create_journey_template	Write	478	2.9%
create_brand	Write	353	2.1%
replace_journey_template	Write	326	2.0%
list_messages	Read	318	1.9%
update_tenant_preference	Write	311	1.9%
create_journey	Write	304	1.8%
create_notification	Write	284	1.7%
put_notification_content	Write	269	1.6%
put_notification_locale	Write	264	1.6%
bulk_add_user_tokens	Write	259	1.6%
replace_routing_strategy	Write	254	1.5%
replace_journey	Write	231	1.4%
create_routing_strategy	Write	230	1.4%
put_notification_element	Write	227	1.4%
send_message_template	Write	219	1.3%
send_message	Write	203	1.2%
invoke_automation_template	Execute	193	1.2%
replace_notification	Write	193	1.2%
create_or_update_tenant	Write	190	1.2%
patch_profile	Write	189	1.1%
invoke_journey	Execute	187	1.1%
create_or_replace_user_push_token	Write	181	1.1%
create_provider	Write	180	1.1%
patch_user_token	Write	179	1.1%
send_message_to_list	Write	179	1.1%
replace_tenant_template	Write	178	1.1%
update_notification_checks	Write	177	1.1%
generate_jwt_for_user	Write	176	1.1%
update_user_preference_topic	Write	170	1.0%
send_message_to_list_template	Write	168	1.0%
update_provider	Write	164	1.0%
update_brand	Write	163	1.0%
track_inbound_event	Execute	157	1.0%
bulk_add_user_tenants	Write	148	0.9%
publish_notification	Write	147	0.9%
courier_installation_guide	Read	146	0.9%
get_journey_template	Read	145	0.9%
subscribe_user_to_lists	Read	140	0.8%
publish_journey_template	Write	138	0.8%
subscribe_user_to_list	Read	137	0.8%
publish_journey	Write	137	0.8%
list_journey_templates	Read	130	0.8%
cancel_automation	Destructive	126	0.8%
create_or_merge_user	Write	124	0.8%
bulk_subscribe_to_list	Read	123	0.7%
get_tenant_template_version	Read	123	0.7%
get_journey	Read	122	0.7%
add_subscribers_to_list	Write	122	0.7%
list_routing_strategy_notifications	Read	121	0.7%
list_journeys	Read	117	0.7%
publish_tenant_template	Write	112	0.7%
get_translation	Read	110	0.7%
update_translation	Write	109	0.7%
delete_tenant_template	Destructive	107	0.6%
update_audience	Write	107	0.6%
list_notification_versions	Read	106	0.6%
list_provider_catalog	Read	105	0.6%
get_notification	Read	104	0.6%
get_user_preference_topic	Read	104	0.6%
add_user_to_tenant	Write	104	0.6%
list_tenant_templates	Read	102	0.6%
replace_profile	Write	102	0.6%
list_automations	Read	101	0.6%
list_journey_template_versions	Read	101	0.6%
list_tenant_users	Read	101	0.6%
get_message_history	Read	98	0.6%
archive_journey_template	Write	98	0.6%
add_bulk_users	Write	97	0.6%
list_routing_strategies	Read	93	0.6%
list_user_tenants	Read	93	0.6%
create_bulk_job	Write	92	0.6%
get_user_preferences	Read	90	0.5%
delete_tenant_preference	Destructive	89	0.5%
list_lists	Read	89	0.5%
list_notification_checks	Read	87	0.5%
get_routing_strategy	Read	86	0.5%
get_tenant_template	Read	86	0.5%
unsubscribe_user_from_list	Read	86	0.5%
archive_routing_strategy	Write	85	0.5%
cancel_notification_submission	Destructive	84	0.5%
delete_user_token	Destructive	84	0.5%
remove_user_from_tenant	Destructive	84	0.5%
get_user_push_token	Read	84	0.5%
get_user_profile_by_id	Read	83	0.5%
create_list	Write	83	0.5%
get_user_list_subscriptions	Read	80	0.5%
list_audience_members	Read	80	0.5%
delete_provider	Destructive	79	0.5%
list_bulk_users	Read	79	0.5%
archive_journey	Write	79	0.5%
get_list_subscribers	Read	78	0.5%
archive_request	Write	78	0.5%
cancel_message	Destructive	76	0.5%
get_message_content	Read	75	0.5%
get_notification_draft_content	Read	75	0.5%
list_journey_versions	Read	74	0.4%
run_bulk_job	Execute	73	0.4%
get_message	Read	73	0.4%
get_audience	Read	72	0.4%
get_audit_event	Read	71	0.4%
list_tenants	Read	71	0.4%
delete_audience	Destructive	69	0.4%
get_notification_content	Read	69	0.4%
delete_user_list_subscriptions	Destructive	68	0.4%
list_user_push_tokens	Read	68	0.4%
delete_tenant	Destructive	67	0.4%
remove_all_user_tenants	Destructive	67	0.4%
get_bulk_job	Read	67	0.4%
archive_notification	Write	67	0.4%
delete_brand	Destructive	66	0.4%
get_provider	Read	66	0.4%
list_audit_events	Read	66	0.4%
delete_profile	Destructive	65	0.4%
get_tenant	Read	65	0.4%
delete_list	Destructive	64	0.4%
get_brand	Read	64	0.4%
get_list	Read	64	0.4%
restore_list	Write	63	0.4%
list_notifications	Read	59	0.4%
list_audiences	Read	58	0.4%
list_providers	Read	58	0.4%
list_brands	Read	57	0.3%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 123.

A PolicyLayer grant exposes only the tools you allow — ungranted definitions are filtered out of the tool list, so they never enter the context window. Estimates below assume typical-weight tools (134 tokens each).

Grant scope	Definition cost	Reduction
All 123 tools (no gateway)	16,505 tokens	—
3 granted tools	~403 tokens	−98%
5 granted tools	~671 tokens	−96%
10 granted tools	~1,342 tokens	−92%

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

//FAQ

Courier token-cost questions.

How many tokens does the Courier MCP server use?+

Its 123 tool definitions total 16,505 tokens — 8.3% 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 Courier 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 Courier's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Courier 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 403 tokens, a 98% 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.

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.

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.

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 07-06-2026 from the PolicyLayer scan database over all 123 catalogued Courier tools. Counts refresh with every site build.

Expose only the tools you use — the rest never enter your context.

A PolicyLayer grant scopes Courier to the tools you actually allow. Ungranted definitions never load, and every call that does run is checked against policy first.

SCOPE YOUR GRANT →

Free to start. No card required.

4,600+ MCP servers and 31,000+ tools scanned and risk-classified.