// MCP TOKEN COST

The Lichess Integration MCP server costs 5,865 tokens before the first call.

Connect Lichess Integration and its 88 tool definitions are loaded into the model's context on every request — 2.9% of a 200k window spent before your agent does anything.

QUICK ANSWER The Lichess Integration MCP server's tool definitions consume 5,865 tokens — 3.1× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 88 tools · 5,865 tokens · 2.9% of 200k · 0.6% 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 2.9%

1M WINDOW 0.6%

Corpus context: Lichess Integration ranks #971 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 5,865 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 67 tokens per tool.

Tool	Category	Tokens	% of server
export_user_games	Write	429	7.3%
create_arena	Write	398	6.8%
create_swiss	Write	230	3.9%
create_challenge	Write	205	3.5%
export_game	Write	199	3.4%
create_simul	Write	196	3.3%
export_games_by_ids	Write	162	2.8%
export_ongoing_game	Write	162	2.8%
get_leaderboard	Read	122	2.1%
get_user_performance	Read	104	1.8%
get_tv_game	Read	101	1.7%
get_users_status	Read	100	1.7%
make_move	Write	88	1.5%
make_board_move	Write	87	1.5%
decline_challenge	Read	86	1.5%
write_in_chat	Write	76	1.3%
push_broadcast_round_pgn	Write	75	1.3%
get_cloud_eval	Read	74	1.3%
get_timeline	Read	70	1.2%
get_ongoing_games	Read	64	1.1%
handle_draw_board_game	Write	63	1.1%
get_puzzle_storm_dashboard	Read	61	1.0%
get_user_profile	Read	60	1.0%
get_user_public_data	Read	60	1.0%
get_puzzle_dashboard	Read	58	1.0%
get_team_members	Read	58	1.0%
decline_join_request	Read	56	1.0%
get_broadcast_round	Read	56	1.0%
search_teams	Read	56	1.0%
get_puzzle_activity	Read	55	0.9%
add_user_note	Write	55	0.9%
export_study_chapter	Write	55	0.9%
accept_join_request	Read	54	0.9%
kick_user_from_team	Read	54	0.9%
send_message	Write	51	0.9%
join_team	Read	50	0.9%
get_users_by_id	Read	48	0.8%
export_all_study_chapters	Write	45	0.8%
test_tokens	Read	44	0.8%
get_puzzle_race	Read	43	0.7%
get_rating_history	Read	43	0.7%
get_team_battle_results	Read	42	0.7%
search_fide_players	Read	42	0.7%
upgrade_to_bot	Destructive	41	0.7%
claim_victory	Read	41	0.7%
get_swiss_games	Read	41	0.7%
get_swiss_info	Read	41	0.7%
get_swiss_results	Read	41	0.7%
set_token	Write	41	0.7%
cancel_challenge	Destructive	40	0.7%
withdraw_from_simul	Financial	40	0.7%
withdraw_from_swiss	Financial	40	0.7%
accept_challenge	Read	40	0.7%
get_arena_games	Read	40	0.7%
get_arena_info	Read	40	0.7%
get_arena_results	Read	40	0.7%
get_broadcast	Read	40	0.7%
get_fide_player	Read	40	0.7%
get_team_join_requests	Read	40	0.7%
get_user_studies	Read	40	0.7%
unfollow_user	Read	40	0.7%
withdraw_from_arena	Financial	39	0.7%
get_thread	Read	39	0.7%
get_user_activity	Read	39	0.7%
resign_board_game	Read	39	0.7%
unblock_user	Read	39	0.7%
set_kid_mode	Write	39	0.7%
get_team_info	Read	38	0.6%
join_simul	Read	38	0.6%
join_swiss	Read	38	0.6%
abort_board_game	Read	37	0.6%
block_user	Read	37	0.6%
follow_user	Read	37	0.6%
join_arena	Read	37	0.6%
get_all_top_10	Read	35	0.6%
leave_team	Read	35	0.6%
get_official_broadcasts	Read	32	0.5%
get_following	Read	30	0.5%
get_tv_channels	Read	30	0.5%
get_my_profile	Read	29	0.5%
revoke_token	Destructive	28	0.5%
get_current_simuls	Read	28	0.5%
create_puzzle_race	Write	28	0.5%
get_arena_tournaments	Read	27	0.5%
get_kid_mode	Read	27	0.5%
list_challenges	Read	27	0.5%
get_my_email	Read	26	0.4%
get_preferences	Read	24	0.4%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 88.

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 (67 tokens each).

Grant scope	Definition cost	Reduction
All 88 tools (no gateway)	5,865 tokens	—
3 granted tools	~200 tokens	−97%
5 granted tools	~333 tokens	−94%
10 granted tools	~666 tokens	−89%

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

//FAQ

Lichess Integration token-cost questions.

How many tokens does the Lichess Integration MCP server use?+

Its 88 tool definitions total 5,865 tokens — 2.9% 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 Lichess Integration 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 Lichess Integration's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Lichess Integration 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 200 tokens, a 97% 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 88 catalogued Lichess Integration tools. Counts refresh with every site build.