// MCP TOKEN COST

The Keploy MCP server costs 39,876 tokens before the first call.

Connect Keploy and its 103 tool definitions are loaded into the model's context on every request — 20% of a 200k window spent before your agent does anything.

QUICK ANSWER The Keploy MCP server's tool definitions consume 39,876 tokens — 21× the median MCP server (1,905 tokens). A scoped grant exposing only the tools you use cuts that roughly in proportion.

MEASURED FROM SCHEMAS 103 tools · 39,876 tokens · 20% of 200k · 4.0% 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 20%

1M WINDOW 4.0%

Corpus context: Keploy ranks #14 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 39,876 tokens go.

Each row is one tool definition as a tools/list entry — name, description and input schema — counted with o200k_base. Average: 387 tokens per tool.

Tool	Category	Tokens	% of server
devloop_scaffold_ci	Execute	2,820	7.1%
replay_sandbox_test	Read	2,521	6.3%
record_sandbox_test	Read	2,373	6.0%
create_test_suite	Write	2,155	5.4%
scaffold_pipeline_workflow	Execute	2,098	5.3%
get_session_report	Read	2,059	5.2%
replay_test_suite	Read	2,002	5.0%
update_test_suite	Write	1,845	4.6%
update_app_custom_variables	Write	1,274	3.2%
devloop_record_sandbox	Execute	1,212	3.0%
devloop_resolve_storage	Write	1,152	2.9%
delete_test_suite	Destructive	952	2.4%
devloop_mutation_demo	Destructive	653	1.6%
updateApp	Write	617	1.5%
get_app_testing_context	Read	568	1.4%
upload_recording	Write	531	1.3%
devloop_generate_resource_flow	Write	513	1.3%
runTestSuites	Execute	473	1.2%
createMock	Write	469	1.2%
updateTestCase	Write	449	1.1%
devloop_expand_coverage	Execute	434	1.1%
devloop_self_heal	Execute	395	1.0%
create_mock	Write	374	0.9%
devloop_detect_app	Read	373	0.9%
getTestReportFull	Read	368	0.9%
devloop_install_repo_artifacts	Write	362	0.9%
updateMock	Write	339	0.9%
uploadRecordingBundle	Write	338	0.8%
create_branch	Write	312	0.8%
download_recording	Read	310	0.8%
editMockMapping	Write	303	0.8%
devloop_switch_to_sandbox	Write	279	0.7%
listTestReports	Read	260	0.7%
update_mock	Write	256	0.6%
devloop_set_storage_mode	Write	250	0.6%
link_mock	Read	247	0.6%
devloop_begin_oauth_install	Write	242	0.6%
getMockMapping	Read	234	0.6%
generateTestSuites	Write	234	0.6%
getTestCase	Read	230	0.6%
list_branches	Read	228	0.6%
getMock	Read	227	0.6%
createApp	Write	224	0.6%
listTestSuites	Read	219	0.5%
run_and_report	Execute	201	0.5%
importRecording	Write	195	0.5%
deleteMock	Destructive	187	0.5%
listMocks	Read	181	0.5%
delete_mock	Destructive	180	0.5%
unlink_mock	Read	178	0.4%
devloop_schema_drift_report	Execute	173	0.4%
listTestCaseReports	Read	170	0.4%
listTestRuns	Read	169	0.4%
delete_recording	Destructive	167	0.4%
listApps	Read	152	0.4%
validateTestSuite	Read	150	0.4%
createCIBranch	Write	148	0.4%
exportRecording	Write	145	0.4%
listSuiteReports	Read	140	0.4%
getTestReport	Read	138	0.3%
getRecording	Read	133	0.3%
listTestSetReports	Read	133	0.3%
listAppsWithRecordings	Read	132	0.3%
generate_and_wait	Write	127	0.3%
normalizeSuiteReport	Write	113	0.3%
getSuiteReport	Read	112	0.3%
listBranches	Read	110	0.3%
deleteTestSuite	Destructive	102	0.3%
getTestSuite	Read	102	0.3%
updateTestSuite	Write	102	0.3%
listRecordings	Read	101	0.3%
devloop_setup_instructions	Write	100	0.3%
streamLoadTestEvents	Read	98	0.2%
stopLoadTest	Execute	87	0.2%
getLoadTestReport	Read	87	0.2%
normalizeTestRun	Read	87	0.2%
getGenerationHistoryDetails	Read	86	0.2%
getTestRun	Read	86	0.2%
bulkDeleteTestSuites	Destructive	81	0.2%
get_auth_status	Read	81	0.2%
listAppsByCluster	Read	81	0.2%
get_setup_instructions	Read	79	0.2%
createTestSuite	Write	79	0.2%
createAPIKey	Write	77	0.2%
getGeneratedSchema	Read	75	0.2%
streamJobEvents	Read	71	0.2%
listLoadTestRuns	Read	70	0.2%
listGenerationHistory	Read	68	0.2%
revokeAPIKey	Destructive	64	0.2%
startLoadTest	Execute	62	0.2%
getSchemaCoverage	Read	62	0.2%
getValidationResult	Read	62	0.2%
stopJob	Execute	60	0.2%
listJobs	Read	59	0.1%
deleteApp	Destructive	58	0.1%
getApp	Read	58	0.1%
getJob	Read	58	0.1%
get_coverage_gaps	Read	51	0.1%
getCurrentUser	Read	40	0.1%
listClusters	Read	37	0.1%
listAPIKeys	Read	33	0.1%
getSubscription	Read	32	0.1%
getUsage	Read	32	0.1%

// SCOPED GRANT

Most agents use a handful of these tools. They pay for all 103.

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 (387 tokens each).

Grant scope	Definition cost	Reduction
All 103 tools (no gateway)	39,876 tokens	—
3 granted tools	~1,161 tokens	−97%
5 granted tools	~1,936 tokens	−95%
10 granted tools	~3,871 tokens	−90%

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

//FAQ

Keploy token-cost questions.

How many tokens does the Keploy MCP server use?+

Its 103 tool definitions total 39,876 tokens — 20% 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 Keploy 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 Keploy's token usage?+

Expose fewer tools. A PolicyLayer grant scopes Keploy 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 1,161 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 103 catalogued Keploy tools. Counts refresh with every site build.