// MCP TOKEN COST

The Agentled MCP server costs 35,543 tokens before the first call.

Connect Agentled and its 119 tool definitions are loaded into the model's context on every request — 18% of a 200k window spent before your agent does anything.

QUICK ANSWER The Agentled MCP server's tool definitions consume 35,543 tokens — 33× the median MCP server (1,075 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 119 tools · 35,543 tokens · 18% of 200k · 3.6% 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 18%

1M WINDOW 3.6%

Corpus context: Agentled ranks #11 of 1,659 measured MCP servers by definition cost. The median is 1,075 tokens, p90 is 6,119, and the heaviest (Fusionauth) is 183,337 — 92% of a 200k window on its own.

// PER-TOOL BREAKDOWN

Where the 35,543 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 299 tokens per tool.

Tool	Category	Tokens	% of server
create_workflow	Write	3,003	8.4%
update_workflow_context	Write	2,453	6.9%
add_step	Write	1,731	4.9%
update_step	Write	1,457	4.1%
update_workflow	Write	1,301	3.7%
update_agent	Destructive	1,276	3.6%
create_agent	Write	961	2.7%
get_step_schema	Read	927	2.6%
patch_timeline_fields	Write	709	2.0%
get_step	Read	692	1.9%
patch_execution_fields	Write	649	1.8%
create_knowledge_list	Write	615	1.7%
upsert_knowledge_rows	Write	530	1.5%
update_knowledge_list_schema	Write	479	1.3%
delete_workflow	Destructive	429	1.2%
create_routine	Write	365	1.0%
move_step	Write	358	1.0%
get_workspace_credits	Read	357	1.0%
start_workflow	Execute	349	1.0%
restore_knowledge_list_snapshot	Write	347	1.0%
configure_channel	Write	338	1.0%
get_knowledge_rows	Read	334	0.9%
update_branding	Write	331	0.9%
submit_feedback_to_agentled	Write	326	0.9%
read_step_output	Read	306	0.9%
get_knowledge_rows_by_ids	Read	304	0.9%
store_memory	Read	304	0.9%
delete_knowledge_rows	Destructive	294	0.8%
list_apps	Read	290	0.8%
list_tracking_events	Read	284	0.8%
list_timelines	Read	277	0.8%
query_kg_edges	Read	276	0.8%
chat	Execute	265	0.7%
validate_workflow	Read	265	0.7%
test_code_action	Read	264	0.7%
test_ai_action	Read	261	0.7%
update_public_form_link	Write	257	0.7%
test_app_action	Read	254	0.7%
upload_agent_file	Write	249	0.7%
rerun_step	Read	247	0.7%
get_workflow_credits	Read	244	0.7%
create_public_form_link	Write	234	0.7%
set_output_page_pin	Write	234	0.7%
create_proactive_agent	Write	228	0.6%
update_routine	Write	225	0.6%
update_workspace_company_profile	Write	225	0.6%
do	Execute	223	0.6%
set_channel_default_agent	Write	217	0.6%
get_workspace_credit_cost_drivers	Read	212	0.6%
search_memories	Read	206	0.6%
snapshot_knowledge_list	Read	206	0.6%
get_workflow_analytics	Read	205	0.6%
manage_agent_workflows	Write	205	0.6%
set_channel_defaults	Write	205	0.6%
preview_n8n_import	Write	199	0.6%
import_n8n_workflow	Write	198	0.6%
chat_with_agent	Write	196	0.6%
list_executions	Read	192	0.5%
rerun	Read	191	0.5%
get_execution	Read	185	0.5%
recall_memory	Read	185	0.5%
upsert_knowledge_text	Write	181	0.5%
list_snapshots	Read	179	0.5%
get_scoring_history	Read	178	0.5%
retry_execution	Read	178	0.5%
import_workflow	Write	176	0.5%
get_snapshot_content	Read	175	0.5%
update_workspace_executive_summary	Write	170	0.5%
remove_step	Destructive	168	0.5%
get_timeline	Read	168	0.5%
list_knowledge_lists	Read	160	0.5%
create_snapshot	Write	157	0.4%
get_knowledge_text	Read	156	0.4%
list_channels	Read	149	0.4%
list_connections	Read	141	0.4%
list_memories	Read	138	0.4%
export_workflow	Write	137	0.4%
publish_workflow	Write	134	0.4%
restore_snapshot	Write	130	0.4%
stop_execution	Execute	120	0.3%
get_app_actions	Read	117	0.3%
list_models	Read	115	0.3%
list_routines	Read	114	0.3%
delete_knowledge_list	Destructive	113	0.3%
get_draft	Read	112	0.3%
delete_snapshot	Destructive	110	0.3%
promote_draft	Write	109	0.3%
update_proactive_agent	Write	103	0.3%
delete_memory	Destructive	100	0.3%
list_agents	Read	100	0.3%
share_execution	Write	99	0.3%
get_workflow	Read	97	0.3%
list_workflows	Read	96	0.3%
discard_draft	Destructive	91	0.3%
get_agent_file	Read	91	0.3%
get_workspace	Read	91	0.3%
delete_agent_file	Destructive	90	0.3%
get_branding	Read	89	0.3%
list_proactive_agents	Read	88	0.2%
activate_agent	Write	87	0.2%
delete_knowledge_text	Destructive	86	0.2%
list_pinned_outputs	Read	86	0.2%
list_agent_files	Read	85	0.2%
resume_routine	Write	85	0.2%
get_workspace_company_profile	Read	84	0.2%
delete_public_form_link	Destructive	82	0.2%
list_public_form_links	Read	79	0.2%
get_proactive_agent	Read	76	0.2%
pause_agent	Read	76	0.2%
pause_routine	Execute	73	0.2%
delete_agent	Destructive	72	0.2%
get_agent	Read	71	0.2%
get_public_form_link	Read	71	0.2%
delete_routine	Destructive	69	0.2%
pause_proactive_agent	Read	68	0.2%
delete_proactive_agent	Destructive	64	0.2%
resume_proactive_agent	Write	62	0.2%
revoke_share	Destructive	60	0.2%
get_share	Read	58	0.2%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 119.

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 (299 tokens each).

Grant scope	Definition cost	Reduction
All 119 tools (no gateway)	35,543 tokens	—
3 granted tools	~896 tokens	−97%
5 granted tools	~1,493 tokens	−96%
10 granted tools	~2,987 tokens	−92%

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

//FAQ

Agentled token-cost questions.

How many tokens does the Agentled MCP server use?+

Its 119 tool definitions total 35,543 tokens — 18% 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 Agentled 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 Agentled's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Agentled 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 896 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 05-06-2026 from the PolicyLayer scan database over all 119 catalogued Agentled tools. Counts refresh with every site build.