// MCP TOKEN COST

The agentView MCP server costs 16,183 tokens before the first call.

Connect agentView and its 81 tool definitions are loaded into the model's context on every request — 8.1% of a 200k window spent before your agent does anything.

QUICK ANSWER The agentView MCP server's tool definitions consume 16,183 tokens — 8.5× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 81 tools · 16,183 tokens · 8.1% of 200k · 1.6% 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 8.1%

1M WINDOW 1.6%

Corpus context: agentView ranks #82 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 16,183 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 200 tokens per tool.

Tool	Category	Tokens	% of server
create_api_key	Write	641	4.0%
send_store_template_to_display	Write	512	3.2%
send_html	Write	506	3.1%
set_data_slot	Write	431	2.7%
search_store_templates	Read	344	2.1%
configure_display	Write	339	2.1%
pair_by_code	Write	324	2.0%
send_url	Write	317	2.0%
get_display_agent_artifact	Read	310	1.9%
get_display_preview_url	Read	298	1.8%
search_public_apis	Read	294	1.8%
list_displays	Read	287	1.8%
create_display	Write	280	1.7%
get_store_template_details	Read	269	1.7%
set_idle_content	Write	269	1.7%
set_display_embeddable_origins	Write	266	1.6%
get_display	Read	265	1.6%
test_display_content	Read	265	1.6%
submit_feedback	Write	264	1.6%
search	Read	258	1.6%
broadcast_content	Write	255	1.6%
claim_display	Destructive	247	1.5%
authenticate	Write	230	1.4%
get_store_template_install_options	Read	228	1.4%
create_auth_session	Write	227	1.4%
invite_member	Write	224	1.4%
get_data_slot_usage	Read	222	1.4%
get_auth_session	Read	207	1.3%
upload_asset	Write	204	1.3%
broadcast_to_categories	Write	202	1.2%
clear_display	Destructive	197	1.2%
create_org_display	Write	197	1.2%
get_store_template_content	Read	195	1.2%
get_display_capabilities	Read	194	1.2%
lock_display	Write	194	1.2%
get_store_template_agent_artifact	Read	193	1.2%
bulk_assign_display_category	Write	189	1.2%
rename_display	Write	186	1.1%
read_display_html	Read	185	1.1%
list_assets	Read	183	1.1%
unlock_display	Write	183	1.1%
set_display_grant	Write	182	1.1%
fetch	Read	181	1.1%
get_account	Read	179	1.1%
set_org_connectivity	Write	175	1.1%
delete_display	Destructive	174	1.1%
get_organization	Read	173	1.1%
set_display_privacy_mode	Write	173	1.1%
set_display_categories_for_display	Write	172	1.1%
get_storage_quota	Read	167	1.0%
list_data_slots	Read	162	1.0%
get_display_content	Read	153	0.9%
allocate_licenses	Read	151	0.9%
create_display_category	Write	149	0.9%
update_member_role	Write	145	0.9%
rename_display_category	Write	142	0.9%
list_organizations	Read	140	0.9%
get_data_slot	Read	134	0.8%
list_store_categories	Read	134	0.8%
create_organization	Write	133	0.8%
assign_license	Write	128	0.8%
delete_data_slot	Destructive	127	0.8%
update_asset	Write	124	0.8%
remove_member	Destructive	123	0.8%
remove_display_grant	Destructive	120	0.7%
remove_display_from_org	Destructive	109	0.7%
unassign_license	Read	107	0.7%
get_billing_url	Read	106	0.7%
get_asset	Read	102	0.6%
list_display_categories	Read	101	0.6%
rename_organization	Write	94	0.6%
list_org_displays	Read	93	0.6%
revoke_api_key	Destructive	91	0.6%
delete_asset	Destructive	90	0.6%
delete_organization	Destructive	89	0.5%
list_public_api_categories	Read	87	0.5%
logout	Destructive	86	0.5%
get_pricing	Read	86	0.5%
get_license_info	Read	80	0.5%
get_public_status	Read	73	0.5%
list_api_keys	Read	67	0.4%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 81.

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 (200 tokens each).

Grant scope	Definition cost	Reduction
All 81 tools (no gateway)	16,183 tokens	—
3 granted tools	~599 tokens	−96%
5 granted tools	~999 tokens	−94%
10 granted tools	~1,998 tokens	−88%

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

//FAQ

agentView token-cost questions.

How many tokens does the agentView MCP server use?+

Its 81 tool definitions total 16,183 tokens — 8.1% 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 agentView 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 agentView's token usage?+

Expose fewer tools. A PolicyLayer grant scopes agentView 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 599 tokens, a 96% 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 07-06-2026 from the PolicyLayer scan database over all 81 catalogued agentView tools. Counts refresh with every site build.