Imported from gabrielmoreira/agent-skills-mirror (mirrors/repos/jeremylongshore@claude-code-plugins-plus-skills/skills/.curated/snowflake-cost-leak-hunter/SKILL.md). Install upstream with npx skills add gabrielmoreira/agent-skills-mirror --skill snowflake-cost-leak-hunter. Copyright stays with the author (MIT).

Snowflake Cost Leak Hunter

Overview

Produce a read-only, evidence-first Snowflake cost investigation. The deterministic analyzer emits a typed ledger whose total, attribution, context, estimate, and invoice-only roles cannot silently collapse into one savings claim.

Problem: Cost views answer different questions, arrive at different times, and do not equal invoice truth. Generic advice easily turns observed credits into unsupported prices or promised savings.

Outcome: A reproducible review packet that identifies what the supplied evidence confirms, what was estimated from an approved rate, what remains at risk, and what is unknown.

Prerequisites

• An exact account and UTC analysis window.
• Sanitized output collected by an operator through an approved read-only Snowflake session, or an equivalent exported evidence bundle.
• A role already authorized to read the required SNOWFLAKE.ACCOUNT_USAGE surfaces.
• A writable local working directory for report artifacts. Use Write only to create new local evidence/report files; never use it to alter Snowflake configuration.
• A customer-supplied rate-card record if currency estimates are requested.

Safety and evidence contract

• Read-only Snowflake work only. Do not execute DDL/DML, change warehouse state or size, alter auto-suspend, assign monitors, create budgets, change tags, or cancel queries. Emit proposed changes for a named owner to review.
• Do not invoke Snowflake authentication from this skill. The operator runs the bounded collection queries through an approved read-only session and supplies only sanitized results. Never request connection files, private keys, tokens, passwords, or environment values.
• Do not require ACCOUNTADMIN. Use a role already authorized to read the needed SNOWFLAKE.ACCOUNT_USAGE views. Visibility differs by database role and account configuration; report missing access rather than escalating privileges.
• Bind every history query to an explicit UTC half-open start and end time. Record session UTC offset, account/organization identity, role, query IDs, collection time, SQL and result hashes, row cap state, and the maximum activity timestamp returned. A single bundle may cover at most seven days; partition longer audits.
• Use the fixed official latency matrix. Apply the reviewed per-view and per-field cutoffs in references/cost-ledger-and-surfaces.md; caller-supplied latency or maximum activity time cannot make a recent window settled. Compute each cutoff from the same-statement execution_context.observed_at, never the later CLI completion timestamp. A quiet source with no recent activity is not thereby stale.
• Credits are not invoice truth. Keep hourly operational credits, daily cloud-adjusted billed credits, and organization-currency/usage-statement evidence as three distinct levels. Resource monitors cover warehouses, including Adaptive Warehouses, but do not cover serverless features or AI services. The issued invoice remains authoritative. See Snowflake's billing reconciliation guide.
• No public price assumptions. Convert credits to currency only when the user supplies an applicable contract/rate-card record. Such conversion remains estimated until reconciled to the billing statement.

Read references/attribution-and-staleness.md before collecting evidence. Read references/warehouse-and-idle-evidence.md for the bounded SQL surfaces. If controls are requested, read references/controls-boundaries.md, but return a review packet only. For Adaptive, storage, transfer, AI, surface-denominator, and typed-ledger rules, read references/cost-ledger-and-surfaces.md.

For a live, model-neutral collection, use the shared read-only collector with an existing Snowflake CLI profile:

python3 "${CLAUDE_SKILL_DIR}/scripts/collect_snowflake_evidence.py" \
  --surface cost --connection <approved-readonly-profile> \
  --window-start <YYYY-MM-DDTHH:MM:SSZ> \
  --window-end <YYYY-MM-DDTHH:MM:SSZ> \
  --output /tmp/snowflake-cost-collector.json

Map the baseline receipt's warehouse, query, load, and generic metering datasets into the analyzer schema. The legacy input key serverless_usage means generic METERING_HISTORY; public labels use metering:<SERVICE_TYPE>. Preserve every receipt field and reject out-of-window rows, cap uncertainty, missing integrity fields, and unscoped fingerprints. A self-checksum proves local consistency, not Snowflake origin.

Collect needed supplemental cost-* surfaces with the same bounded collector and put each complete receipt under supplemental_receipts.<surface_inventory_name>. Follow the surface contract for the exact dataset keys, hashes, latency cutoffs, privacy rules, and unavailable-surface behavior.

Instructions

1. Fix scope before querying

Capture:

• account and role;
• half-open UTC window [window_start, window_end);
• requested attribution dimension, such as warehouse, user, query tag, or service;
• whether an approved contract rate card is available;
• whether Adaptive Warehouses or serverless features are in scope.
• the complete expected-surface denominator, including storage, transfer, AI, resource-monitor, and budget evidence when those domains are in scope.

If the user supplies only an invoice total, state that the audit can explain usage but cannot reconcile the invoice without the corresponding usage statement and contract rates.

2. Verify access without changing grants

Have the operator run the smallest read probes with the approved connection and provide the sanitized results. This skill does not expose the Snowflake CLI namespace. A representative probe is:

SELECT MAX(end_time) AS max_end_time
FROM SNOWFLAKE.ACCOUNT_USAGE.WAREHOUSE_METERING_HISTORY
WHERE start_time >= :window_start
  AND start_time < :window_end;

Probe QUERY_ATTRIBUTION_HISTORY separately because its availability, latency, and coverage differ. If either probe fails, preserve the exact sanitized error, name the missing surface, and stop that branch. Do not propose granting broad imported privileges automatically.

3. Collect normalized evidence

Use the queries and field definitions in references/warehouse-and-idle-evidence.md. Export only the normalized fields accepted by scripts/analyze_cost_evidence.py; exclude raw SQL text and credentials.

The input accepts baseline warehouse/query/load/metering arrays; supplemental Adaptive, storage, transfer, AI, monitor, and budget evidence; optional invoice rows; and user-supplied credit rates. Read the surface contract for the exact keys and overlap rules. Inventory assertions without matching receipts block completeness.

Record each source's maximum activity timestamp only as descriptive context; it does not prove freshness, and an absent source is not zero usage. Apply the reference's identity-disclosure contract. Never export raw users, tags, query text, notification addresses, contract numbers, credentials, or connection values.

4. Run deterministic analysis

Print the canonical bundle digest in a trusted local terminal and record it separately from the bundle. Never trust a digest copied from inside the evidence itself:

python3 "${CLAUDE_SKILL_DIR}/scripts/analyze_cost_evidence.py" \
  --input cost-evidence.json --print-input-sha256

Then pass that separately recorded value back to the analyzer. Live receipts older than one hour are rejected as stale transport evidence; recollect rather than relaxing the bound.

python3 "${CLAUDE_SKILL_DIR}/scripts/analyze_cost_evidence.py" \
  --input cost-evidence.json \
  --trusted-input-sha256 <recorded-sha256:hex-digest> \
  --json-out cost-analysis.json \
  --markdown-out cost-analysis.md

The script validates non-negative numeric evidence, sums with decimal arithmetic, and keeps total, attribution, context, estimate, and invoice-only ledger roles separate. Their additive and overlap rules are defined in the surface contract.

It does not apply magnitude thresholds, infer a price, or recommend a warehouse size. Daily average storage snapshots must remain per day or use an explicitly labeled average/byte-day calculation; summing multiple daily snapshots and labeling the result bytes is invalid. When query fingerprints have both attributed credits and elapsed time, it also emits a non-dominance cost/latency Pareto view. A Pareto point is a comparison aid, not a proof that a workload should move warehouses. Right-sizing is only a bounded review proposal when the operator supplies the current size, explicit candidate sizes, maximum size steps, measurement window, success criteria, and approver; never infer a target size from credits or queue time.

5. Corroborate before recommending

For each ranked opportunity, record:

1. the exact source rows and time window;
2. the observed source age;
3. coverage exclusions or NULL fields;
4. a competing explanation;
5. a read-only next measurement;
6. the owner who would approve any later change.

Examples of competing explanations include intentionally warm warehouses, SLA-driven capacity, untagged shared-service queries, or usage outside the attribution view's coverage. Do not label those cases waste without workload-owner confirmation.

6. Deliver the review packet

Follow references/output-contract.md. Lead with the window and coverage, not a sensational savings number. A valid packet contains:

• confirmed credits by evidence surface;
• the typed ledger with parent IDs, overlap keys, aggregation eligibility, freshness, availability, and invoice-reconciliation status;
• attribution completeness by warehouse, including unknown boundaries for NULL attribution and query coverage gaps;
• cost/latency Pareto points by query fingerprint and warehouse-load correlation;
• estimated currency in a separate table, if and only if a rate card was supplied;
• at-risk opportunities ranked by observed credits, each labeled review required;
• missing/late-source warnings;
• read-only verification queries;
• proposed changes in an approval queue, with no execution performed.

Output

Return cost-analysis.json and cost-analysis.md in the user's chosen working directory, plus the exact analyzer command used. The JSON is the machine-readable receipt; Markdown is the human review packet. Both must contain the analysis window, source freshness, confirmed observations, estimated amounts, at-risk opportunities, warnings, and non-claims. Do not write runtime output into the skill directory.

Stop conditions

Stop and return a bounded partial result when:

• authentication or the approved role fails;
• the requested window is newer than the available source timestamps;
• account and organization usage are mixed without aligned account identifiers and UTC boundaries;
• Adaptive Warehouse rows make warehouse attribution columns NULL;
• a required cost surface is unavailable, outside its fixed official settled window, truncated, region-limited, or hidden by the approved role;
• a currency request has no applicable contract rate;
• evidence contains negative credits, malformed timestamps, or incompatible currencies;
• the only proposed next step would mutate production.

Error Handling

Examples

“Why did warehouse credits jump last week?”

Collect the two warehouse surfaces over the same UTC window. The analyzer may report 42.5 confirmed warehouse compute credits and 11.2 credits at risk for idle-time review. It must not call all 11.2 credits waste or convert them to dollars without a supplied rate.

“Show costs by team from QUERY_TAG”

Aggregate QUERY_ATTRIBUTION_HISTORY by the existing tag. Report tagged and untagged credits, the view's maximum timestamp, and exclusions such as idle, serverless, storage, and cloud-services cost. Missing tags are an at-risk attribution gap, not proof of unowned spend.

“Create a monitor to shut down expensive warehouses”

Do not create or alter a monitor. Audit current warehouse evidence, explain the warehouse-only coverage and suspension caveats from references/controls-boundaries.md, and return a reviewable control proposal requiring explicit authorization.

Resources

• Attribution and staleness — view coverage, latency, privilege, and invoice-reconciliation boundaries.
• Warehouse and idle evidence — bounded collection queries and normalization schema.
• Controls boundaries — resource-monitor and budget semantics, including uncovered serverless usage.
• Pareto and right-sizing — cost/latency frontier and bounded, one-variable resize review.
• Output contract — evidence labels and review-packet format.
• scripts/analyze_cost_evidence.py — deterministic validator, classifier, and renderer.