// MCP TOKEN COST

The VaultCrux Memory Core MCP server costs 14,412 tokens before the first call.

Connect VaultCrux Memory Core and its 76 tool definitions are loaded into the model's context on every request — 7.2% of a 200k window spent before your agent does anything.

QUICK ANSWER The VaultCrux Memory Core MCP server's tool definitions consume 14,412 tokens — 7.6× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 76 tools · 14,412 tokens · 7.2% of 200k · 1.4% 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 7.2%

1M WINDOW 1.4%

Corpus context: VaultCrux Memory Core ranks #97 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 14,412 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 190 tokens per tool.

Tool	Category	Tokens	% of server
cuecrux_session	Read	583	4.0%
memory_retrieve	Read	386	2.7%
memory_reason_about	Read	358	2.5%
declare_revenue_willingness	Read	325	2.3%
query_vault	Read	300	2.1%
action_journal_query	Read	284	2.0%
create_work	Write	244	1.7%
proof_document	Read	234	1.6%
get_beliefs	Read	229	1.6%
get_journal	Read	228	1.6%
get_watch_alerts	Read	228	1.6%
memory_engrams_resolve	Write	227	1.6%
create_coalition	Write	222	1.5%
register_belief	Write	220	1.5%
query_with_threshold	Read	218	1.5%
submit_feature_request	Write	218	1.5%
register_agent	Write	213	1.5%
update_work_state	Write	213	1.5%
forecast_obsolescence	Read	210	1.5%
get_feature_requests	Read	207	1.4%
get_knowledge_gaps	Read	201	1.4%
schedule_recheck	Write	198	1.4%
tip_agent	Write	194	1.3%
memory_session_init	Read	193	1.3%
list_work	Read	192	1.3%
set_policy	Write	187	1.3%
get_stale_pins	Read	186	1.3%
watch_answer	Read	184	1.3%
pin_receipt	Write	184	1.3%
vote_feature_request	Write	184	1.3%
github_search	Read	182	1.3%
join_coalition	Read	180	1.2%
comment_on_work	Write	180	1.2%
annotate_session	Write	178	1.2%
get_blast_radius	Read	177	1.2%
list_seats	Read	177	1.2%
get_break_analysis	Read	176	1.2%
find_contradictions	Read	175	1.2%
get_watches	Read	175	1.2%
create_handoff_package	Write	174	1.2%
get_proof_chunks	Read	172	1.2%
tip_platform	Write	172	1.2%
purchase_bundle	Read	171	1.2%
invite_seat	Write	170	1.2%
set_reasoning_profile	Write	170	1.2%
github_open_issues	Write	169	1.2%
change_seat_role	Read	167	1.2%
get_proofpack	Read	166	1.2%
github_open_prs	Write	166	1.2%
get_proof_receipt	Read	165	1.1%
get_proof_status	Read	165	1.1%
diff_receipts	Read	164	1.1%
request_sponsor	Read	160	1.1%
get_counterfactual_summary	Read	159	1.1%
browse_bundles	Read	158	1.1%
github_recent_commits	Read	154	1.1%
revoke_seat	Destructive	152	1.1%
accept_handoff_package	Read	152	1.1%
unwatch_answer	Destructive	148	1.0%
get_project_context	Read	148	1.0%
get_spend_receipt	Read	148	1.0%
github_comments_since	Read	146	1.0%
explain_last_answer	Read	145	1.0%
verify_passport	Read	143	1.0%
list_projects	Read	139	1.0%
get_economy_dashboard	Read	137	1.0%
get_active_policy	Read	136	0.9%
get_credit_balance	Read	133	0.9%
get_session_context	Read	133	0.9%
get_credit_escrow	Read	127	0.9%
get_trust_level	Read	127	0.9%
get_domain_affinity	Read	126	0.9%
get_reasoning_profile	Read	126	0.9%
get_daily_briefing	Read	125	0.9%
get_pricing	Read	125	0.9%
get_passport	Read	124	0.9%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 76.

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 (190 tokens each).

Grant scope	Definition cost	Reduction
All 76 tools (no gateway)	14,412 tokens	—
3 granted tools	~569 tokens	−96%
5 granted tools	~948 tokens	−93%
10 granted tools	~1,896 tokens	−87%

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

//FAQ

VaultCrux Memory Core token-cost questions.

How many tokens does the VaultCrux Memory Core MCP server use?+

Its 76 tool definitions total 14,412 tokens — 7.2% 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 VaultCrux Memory Core 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 VaultCrux Memory Core's token usage?+

Expose fewer tools. A PolicyLayer grant scopes VaultCrux Memory Core 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 569 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 76 catalogued VaultCrux Memory Core tools. Counts refresh with every site build.