Provides read-only visibility into Snowflake warehouse cost and query performance, including credit usage by warehouse, identification of expensive queries, and heuristic right-sizing recommendations, using Snowflake's ACCOUNT_USAGE views.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@snowflake-cost-mcp Show credit usage by warehouse for last week"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP server that gives Claude (or any MCP-speaking agent) direct, read-only visibility into Snowflake warehouse cost and query performance — credit usage by warehouse, the most expensive queries in a window, and heuristic right-sizing recommendations.
Snowflake already ships an official, actively-developed managed MCP server covering schema browsing, Cortex Analyst/Search, and general query execution — and several community servers cover similar ground. This project intentionally does not duplicate any of that. It exists for the one thing none of them focus on: cost attribution and performance right-sizing, using Snowflake's ACCOUNT_USAGE views.
Use this alongside the official Snowflake MCP server, not instead of it.
Related MCP server: BigQuery FinOps MCP Server
Tool | What it does |
| Warehouse inventory: size, auto-suspend/resume, running/queued queries |
| Credits consumed per warehouse over a lookback window |
| Slowest/costliest successful queries in a window |
| Full detail for one query by |
| Heuristic oversized/undersized/no-signal call per warehouse |
| Ad-hoc |
ACCOUNT_USAGE views are eventually consistent (Snowflake documents up to ~45 minutes–a few hours of latency), so treat results as "recent history," not real-time.
You need a Snowflake role that can read ACCOUNT_USAGE. Create a dedicated, least-privilege role rather than reusing ACCOUNTADMIN:
CREATE ROLE IF NOT EXISTS mcp_cost_monitor;
GRANT IMPORTED PRIVILEGES ON DATABASE snowflake TO ROLE mcp_cost_monitor;
GRANT USAGE ON WAREHOUSE compute_xs TO ROLE mcp_cost_monitor;
GRANT ROLE mcp_cost_monitor TO USER your_service_user;This role can read account usage metadata and nothing else — no write grants anywhere. That's the real security boundary; the in-code SQL guard is a backup, not a substitute for this.
pip install -e .openssl genrsa 2048 | openssl pkcs8 -topk8 -inform PEM -out rsa_key.p8 -nocrypt
openssl rsa -in rsa_key.p8 -pubout -out rsa_key.pubALTER USER your_service_user SET RSA_PUBLIC_KEY='<contents of rsa_key.pub, header/footer stripped>';Copy .env.example to .env and fill in SNOWFLAKE_ACCOUNT, SNOWFLAKE_USER, and SNOWFLAKE_PRIVATE_KEY_PATH (pointing at rsa_key.p8). Username/password auth is supported as a fallback for local/dev use — see .env.example.
snowflake-cost-mcpIt will sit waiting on stdio — that's expected; it's meant to be launched by an MCP client, not run interactively.
{
"mcpServers": {
"snowflake-cost": {
"command": "snowflake-cost-mcp",
"env": {
"SNOWFLAKE_ACCOUNT": "your_account_identifier",
"SNOWFLAKE_USER": "your_service_user",
"SNOWFLAKE_PRIVATE_KEY_PATH": "/absolute/path/to/rsa_key.p8",
"SNOWFLAKE_ROLE": "MCP_COST_MONITOR"
}
}
}
}Every built-in tool runs a hardcoded, parameterized SQL template — no string-built SQL from user input, ever.
run_readonly_query (the only tool that accepts free-form SQL) is off by default and, when enabled, is passed through a guard that rejects anything except a plain SELECT / WITH ... SELECT.
The Snowflake role itself should be read-only (see Prerequisites) — that's the boundary that actually matters if there's ever a bug in this server.
pip install -e ".[dev]"
pytest -q
ruff check src testsTests run entirely against a mocked Snowflake connection — no live account needed to develop or run CI.
Validate end-to-end against a real Snowflake trial account
Publish to the public MCP registry
Add a storage-cost tool (database/table storage spend)
Optional streamable-HTTP transport for a hosted/enterprise mode
MIT — see LICENSE.
Find the slowest / most expensive successful queries in the lookback window, ranked by elapsed time. Each result includes a truncated query text preview, warehouse, user, elapsed seconds, and GB scanned.
Args: lookback_days: How many days back to look (default from server config). limit: Max number of queries to return (default from server config, typically 50).
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | ||
| lookback_days | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations exist, so description must cover behavioral traits. It mentions it returns 'successful queries', ranked by elapsed time, and lists output fields (truncated query text, warehouse, user, etc.). It does not mention auth, rate limits, or potential repercussions, but for a read-only analysis tool, the disclosed behavior is sufficient and not misleading.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise: a single introductory sentence followed by a clear 'Args' block. Every sentence adds value, no redundancy. The information is front-loaded with the primary purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given 2 simple parameters and an output schema (exists but not shown), the description adequately explains the return fields and ranking criteria. It lacks error handling or edge cases, but for a straightforward lookup tool, it is complete enough for an agent to use effectively.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The description includes an 'Args' section that explains both parameters: 'lookback_days' as how many days back to look (with default from server config) and 'limit' as max number of queries (default typically 50). The input schema only provides type and nullability, so the description adds essential meaning and default behavior.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states 'Find the slowest / most expensive successful queries', which is a specific verb ('Find') and resource ('successful queries'). It distinguishes from sibling tools like 'get_query_detail' (specific query) and 'run_readonly_query' (execute a query) by focusing on ranking expensive ones.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description does not explicitly state when to use or not use this tool versus alternatives, but the purpose is clear enough for an agent to infer that it's for finding expensive queries, while siblings serve different purposes (e.g., listing warehouses, getting details). No exclusions or when-not listed, but context is adequate.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Get full detail for a single query by its Snowflake QUERY_ID, including warehouse size, database/schema, elapsed time, bytes scanned, rows produced, and any error message.
Args: query_id: The Snowflake QUERY_ID (as shown in QUERY_HISTORY or the Snowsight UI).
| Name | Required | Description | Default |
|---|---|---|---|
| query_id | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It lists output fields but does not disclose whether the operation is read-only, any side effects, or authorization requirements. The description is adequate but could be more transparent.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise: a single sentence for purpose, a bullet list of output fields, and an Args section for the parameter. Every sentence adds value, and it is front-loaded with the core function.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given that an output schema exists (though not shown), the description adequately lists the main fields returned. It could mention that the operation is read-only or non-destructive, but overall it provides sufficient context for an agent to use the tool correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has 0% description coverage, so the description must compensate. It provides a clear, helpful explanation for the lone parameter query_id: 'The Snowflake QUERY_ID (as shown in QUERY_HISTORY or the Snowsight UI).' This adds significant meaning beyond the schema field name.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it gets full detail for a single query by QUERY_ID, naming specific fields returned. It distinguishes from sibling tools like find_expensive_queries and get_warehouse_credit_usage, which serve different purposes.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies when to use the tool: for detailed info on a specific query. It provides clear context but does not explicitly state when not to use it or mention alternatives.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Total Snowflake credits consumed per warehouse over the lookback window, split into compute vs. cloud services credits. Use this to see where spend is concentrated before drilling into individual warehouses or queries.
Args: lookback_days: How many days back to look (default from server config, typically 7).
| Name | Required | Description | Default |
|---|---|---|---|
| lookback_days | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations, so description carries full burden. It explains what the tool does and what output to expect (total credits per warehouse, split). Does not mention side effects but for a read-only aggregation tool this is sufficient. Could be more explicit about the scope (all warehouses, only active?).
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences plus an Args block. Front-loaded with purpose, no wasted words. Efficiently conveys all necessary information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the simplicity (one param, output schema exists), the description covers purpose, usage context, and parameter. Could briefly note that it returns per-warehouse data, but output schema likely handles details. Missing maybe a note about data freshness or caching.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0% so description must explain parameters. The one parameter 'lookback_days' is clearly described: 'How many days back to look (default from server config, typically 7).' Provides default behavior and typical value.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Clearly states verb 'get' and resource 'warehouse credit usage'. Specifies aggregation per warehouse and split into compute vs cloud services. Distinguishes from siblings like find_expensive_queries or list_warehouses.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides usage context: 'Use this to see where spend is concentrated before drilling into individual warehouses or queries.' Implies a hierarchical analysis workflow. Does not explicitly mention when not to use, but context signals suggest alternatives exist.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Heuristic right-sizing recommendations per warehouse, combining credit usage with query queueing behavior:
HIGH credits + very low query count -> likely oversized or left running idle; check auto-suspend and whether it's needed at all.
Meaningful average queued-overload time -> likely undersized or under-provisioned for concurrency; consider a larger size or a multi-cluster warehouse.
Otherwise -> no strong signal either way.
This is a heuristic starting point for investigation, not a guaranteed cost-saving recommendation -- always sanity-check against actual workload requirements before resizing a production warehouse.
Args: lookback_days: How many days back to look (default from server config).
| Name | Required | Description | Default |
|---|---|---|---|
| lookback_days | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It discloses the heuristic nature, that it's a starting point not guaranteed, and advises caution. It explains the logic behind recommendations but does not detail return format or side effects.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured with an introduction, bullet points for heuristics, and a caution. It is concise but front-loads the key purpose. Slightly verbose in the bullet examples but overall efficient.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the existence of an output schema (not shown), the description covers the tool's purpose, heuristic logic, and parameter meaning. It omits details on the exact output, but the output schema likely handles that. It is complete for a heuristic investigation tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Only one parameter (lookback_days) with 0% schema coverage. Description adds 'default from server config' which provides context beyond the schema, but could specify format or range more clearly.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it provides heuristic right-sizing recommendations per warehouse, combining credit usage and query queueing behavior. It differentiates from sibling tools like get_warehouse_credit_usage and list_warehouses by offering synthesized insights.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides scenarios for when to suspect oversized or undersized warehouses and advises sanity-checking before resizing. However, it does not explicitly state when not to use or mention alternatives among siblings.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all warehouses visible to the connected role, with size, auto-suspend, and auto-resume settings, and current running/queued query counts.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. It indicates a read-only listing operation with specific data, but does not disclose edge cases, rate limits, or behavior under load. Adequate but not exceptional.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Single sentence, 20 words, efficiently conveys all key points. Front-loaded with action and resource, then details. Every word earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no parameters and presence of output schema, description covers scope and key fields. Could mention pagination or default ordering, but for a simple list tool, it is sufficiently complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Input schema has zero parameters, so description does not need to add semantics. Baseline 4 applies; description adds no parameter info but that is unnecessary. It does mention output fields.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states the verb 'list' and resource 'warehouses', specifies exact details returned (size, auto-suspend/resume settings, query counts), and implicitly distinguishes from siblings that deal with queries or recommendations.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Description implies usage for listing warehouses, but lacks explicit guidance on when to use this vs alternatives like get_warehouse_credit_usage or run_readonly_query. No exclusions or prerequisites mentioned.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Run an arbitrary read-only SELECT query against Snowflake, for cost or performance investigation beyond the built-in tools. Disabled by default -- must be explicitly enabled via SNOWFLAKE_MCP_ALLOW_ARBITRARY_QUERIES=true in the server environment. Only SELECT / WITH...SELECT statements are accepted; the connected Snowflake role should also be read-only as the primary security boundary.
Args: sql: A single read-only SELECT statement.
| Name | Required | Description | Default |
|---|---|---|---|
| sql | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Discloses read-only nature, security boundary via role limitation, and enablement requirement. With no annotations, description carries full burden and does well, though rate limits or error handling omitted.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences plus a one-line parameter doc, front-loaded with purpose and constraints. Every sentence adds value.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given one parameter and output schema present, description adequately covers purpose, constraints, enablement, and parameter semantics. Output schema details not needed in description.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema has 0% coverage but description clarifies 'sql' must be a 'single read-only SELECT statement', adding constraint beyond the schema. Compensates for missing schema description.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States 'run an arbitrary read-only SELECT query against Snowflake' with specific purpose 'for cost or performance investigation beyond the built-in tools', clearly distinguishing from sibling tools.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Indicates it is disabled by default and must be explicitly enabled, only accepts SELECT/WITH...SELECT, and advises a read-only role. Could explicitly mention alternatives but implies siblings are for common investigations.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
v0.1.0find_expensive_queriesget_query_detailget_warehouse_credit_usageget_warehouse_right_sizing_recommendationslist_warehousesrun_readonly_queryEach tool targets a distinct aspect of Snowflake cost and performance: queries (expensive vs detail), warehouses (credit usage, sizing recommendations, listing), and a fallback for arbitrary queries. No overlap in purpose.
All tool names follow a consistent verb_noun pattern using snake_case (find_expensive_queries, get_query_detail, list_warehouses, etc.). Predictable and readable.
With 6 tools, the surface is well-scoped for Snowflake cost management—covering query and warehouse analysis without bloat. Each tool earns its place.
Covers core cost investigation (expensive queries, warehouse usage, sizing) and includes a generic run tool for gaps. Missing finer breakdowns (e.g., by user/database) but no critical dead ends.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Query cloud, AI and SaaS spend across 25+ providers: costs, budgets, anomalies, unit economics.
Cloud cost, inventory and governance on AWS/Azure/GCP. Read-only by default, optional scoped writes
Read-only AWS cost analysis: find idle and underutilized resources, with evidence.
AWS Cost Explorer cost, forecast, and anomaly reporting through user-connected IAM credentials.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/alizubairs/Snowflake-Cost-Performance-MCP-server'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
