// MCP TOKEN COST

The Hwpx MCP server costs 19,131 tokens before the first call.

Connect Hwpx and its 135 tool definitions are loaded into the model's context on every request — 9.6% of a 200k window spent before your agent does anything.

QUICK ANSWER The Hwpx MCP server's tool definitions consume 19,131 tokens — 10× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 135 tools · 19,131 tokens · 9.6% of 200k · 1.9% 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 9.6%

1M WINDOW 1.9%

Corpus context: Hwpx ranks #57 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 19,131 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 142 tokens per tool.

Tool	Category	Tokens	% of server
build_document	Execute	1,172	6.1%
render_mermaid	Read	750	3.9%
insert_image	Write	689	3.6%
render_mermaid_in_cell	Write	454	2.4%
insert_image_in_cell	Write	393	2.1%
update_table_cell	Write	322	1.7%
insert_table	Write	320	1.7%
set_text_style	Write	303	1.6%
insert_paragraph	Write	301	1.6%
format_text	Read	292	1.5%
update_paragraph_text_preserve_styles	Write	273	1.4%
batch_fill_table	Write	255	1.3%
update_paragraph_text	Write	255	1.3%
fill_by_path	Write	253	1.3%
replace_text	Write	247	1.3%
find_insert_position_after_header	Read	241	1.3%
set_table_cell_auto_hanging_indent	Write	233	1.2%
get_cell_context	Read	232	1.2%
get_user_paths	Read	226	1.2%
replace_text_in_cell	Write	215	1.1%
insert_nested_table	Write	205	1.1%
set_paragraph_style	Write	198	1.0%
insert_ellipse	Write	192	1.0%
set_numbering	Write	191	1.0%
chunk_document	Read	190	1.0%
get_tool_guide	Read	189	1.0%
insert_rect	Write	187	1.0%
merge_cells	Write	186	1.0%
set_table_cell_hanging_indent	Write	186	1.0%
insert_line	Write	185	1.0%
find_cell_by_label	Read	184	1.0%
get_table_map	Read	180	0.9%
set_hanging_indent	Write	179	0.9%
set_auto_hanging_indent	Write	174	0.9%
search_chunks	Read	172	0.9%
extract_toc	Read	170	0.9%
set_cell_properties	Write	170	0.9%
set_cell_background_color	Write	169	0.9%
set_page_settings	Write	168	0.9%
set_section_xml	Write	154	0.8%
copy_table	Write	150	0.8%
remove_table_cell_hanging_indent	Destructive	149	0.8%
get_chunk_context	Read	149	0.8%
move_table	Write	147	0.8%
get_table_cell_hanging_indent	Read	144	0.8%
split_cell	Read	144	0.8%
build_position_index	Execute	140	0.7%
repair_xml	Read	136	0.7%
set_column_def	Write	135	0.7%
set_column_widths	Write	135	0.7%
insert_table_row	Write	133	0.7%
search_text	Read	125	0.7%
find_insert_position_after_table	Read	123	0.6%
copy_paragraph	Write	123	0.6%
move_paragraph	Write	123	0.6%
set_raw_section_xml	Write	120	0.6%
save_document	Write	117	0.6%
insert_hyperlink	Write	116	0.6%
delete_table_row	Destructive	114	0.6%
search_position_index	Read	114	0.6%
get_insert_context	Read	113	0.6%
get_table_cell	Read	113	0.6%
update_image_size	Write	113	0.6%
set_footer	Write	112	0.6%
set_header	Write	112	0.6%
insert_equation	Write	107	0.6%
get_tables_summary	Read	106	0.6%
insert_memo	Write	106	0.6%
insert_table_column	Write	106	0.6%
insert_page_break	Write	102	0.5%
insert_footnote	Write	101	0.5%
delete_table_column	Destructive	100	0.5%
apply_style	Write	100	0.5%
insert_endnote	Write	100	0.5%
append_text_to_paragraph	Read	99	0.5%
batch_replace	Write	97	0.5%
insert_bookmark	Write	97	0.5%
find_paragraph_by_text	Read	96	0.5%
get_element_index_for_table	Read	94	0.5%
get_table_as_csv	Read	94	0.5%
set_document_metadata	Write	93	0.5%
get_text_style	Read	92	0.5%
get_section_xml	Read	87	0.5%
delete_table	Destructive	84	0.4%
delete_image	Destructive	83	0.4%
remove_hanging_indent	Destructive	83	0.4%
get_chunk_at_offset	Read	83	0.4%
analyze_xml	Read	82	0.4%
delete_paragraph	Destructive	81	0.4%
get_hanging_indent	Read	78	0.4%
get_table	Read	78	0.4%
get_paragraph	Read	77	0.4%
get_raw_section_xml	Read	75	0.4%
find_table_by_header	Read	74	0.4%
get_paragraph_style	Read	74	0.4%
insert_section	Write	71	0.4%
redo	Read	70	0.4%
get_paragraphs	Read	69	0.4%
undo	Read	69	0.4%
delete_memo	Destructive	66	0.3%
get_tables_by_section	Read	66	0.3%
get_page_settings	Read	65	0.3%
get_position_index	Read	65	0.3%
delete_section	Destructive	64	0.3%
invalidate_reading_cache	Destructive	64	0.3%
get_column_def	Read	62	0.3%
get_footer	Read	61	0.3%
get_header	Read	61	0.3%
export_to_text	Write	61	0.3%
export_to_html	Write	60	0.3%
open_document	Write	58	0.3%
find_empty_tables	Read	57	0.3%
get_images	Read	56	0.3%
create_document	Write	56	0.3%
get_document_outline	Read	54	0.3%
get_document_structure	Read	51	0.3%
get_bullet_defs	Read	50	0.3%
get_document_metadata	Read	50	0.3%
get_numbering_defs	Read	50	0.3%
get_footnotes	Read	46	0.2%
get_memos	Read	46	0.2%
get_document_text	Read	45	0.2%
get_endnotes	Read	45	0.2%
get_hyperlinks	Read	45	0.2%
list_template_profiles	Read	45	0.2%
get_bookmarks	Read	44	0.2%
get_equations	Read	44	0.2%
get_styles	Read	44	0.2%
get_word_count	Read	44	0.2%
close_document	Write	44	0.2%
get_char_shapes	Read	43	0.2%
get_para_shapes	Read	43	0.2%
get_sections	Read	43	0.2%
get_tables	Read	43	0.2%
list_open_documents	Read	27	0.1%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 135.

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 (142 tokens each).

Grant scope	Definition cost	Reduction
All 135 tools (no gateway)	19,131 tokens	—
3 granted tools	~425 tokens	−98%
5 granted tools	~709 tokens	−96%
10 granted tools	~1,417 tokens	−93%

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

//FAQ

Hwpx token-cost questions.

How many tokens does the Hwpx MCP server use?+

Its 135 tool definitions total 19,131 tokens — 9.6% 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 Hwpx 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 Hwpx's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Hwpx 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 425 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.

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 135 catalogued Hwpx tools. Counts refresh with every site build.