dbt-projects-on-snowflake — скилл для Claude Code: установка и запуск
Скилл dbt-projects-on-snowflake проводит агента через деплой dbt-проектов внутрь Snowflake: рабочее пространство Snowsight, объекты DBT PROJECT, запуск по расписанию и разбор телеметрии из event-таблиц.
Apache-2.0 1 установок ★ 35 обновлён 2026-10-27
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a claude-codeЧто делает dbt-projects-on-snowflake
Скилл описывает три пути к dbt-проекту, который живёт в Snowflake: рабочее пространство Snowsight, объект DBT PROJECT на уровне схемы и Snowflake CLI. Агент подскажет нужную команду — snow dbt deploy my_project --source . для загрузки, snow dbt execute my_project build для запуска или чистый SQL вроде EXECUTE DBT PROJECT <db>.<schema>.<project> args='build', если удобнее остаться в worksheet. Флаги вроде --full-refresh и --select tag:gold передаются прямо в args.
Версии не зашиты в текст скилла. Список отдаёт SYSTEM$SUPPORTED_DBT_VERSIONS(), а атрибут DBT_VERSION попутно выбирает движок: ниже 2.0 — dbt Core на Python, от 2.0 — dbt Fusion на Rust. Закрепить версию можно на уровне аккаунта, при создании проекта или на один запуск, не меняя определение.
Мониторинг строится на event-таблицах по модели OpenTelemetry, а готовые запросы лежат в scripts/ и запускаются через --enable-templating JINJA -D event_table=.... Внешний доступ, интеграция с Git API и настройка самой event-таблицы вынесены в references/SETUP.md — без этого шага скриптам мониторинга нечего читать. Фильтр по TIMESTAMP ставьте в каждый запрос сразу, иначе просканируется вся таблица и расход credits вырастет.
Какие задачи решает
- развернуть dbt-проект в Snowflake прямо из рабочей папки
- гнать
buildс--full-refreshпо расписанию через задачи Snowflake - закрепить версию dbt за одним проектом, не трогая остальные
- вытащить упавшие запуски из логов event-таблицы
- сравнить потребление CPU и памяти между проектами
Примеры запросов агенту
- Разверни dbt-проект из текущей папки в Snowflake и запусти
build. - Покажи поддерживаемые версии dbt и закрепи
2.0.0-preview.186за проектомmy_dbt_project. - Собери отчёт по ошибкам выполнения из event-таблицы
LOGS_DB.PUBLIC.DBT_EVENTS.
Как установить dbt-projects-on-snowflake
1Через Skills CLI (Vercel)
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a claude-code2Вручную: положить файлы
~/.claude/skills/dbt-projects-on-snowflake/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/dbt-projects-on-snowflake/
1Через Skills CLI (Vercel)
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a codex2Вручную: положить файлы
.agents/skills/dbt-projects-on-snowflake/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a cursor2Вручную: положить файлы
.cursor/skills/dbt-projects-on-snowflake/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install sfc-gh-dflippo/snowflake-dbt-demo .claude/skills/dbt-projects-on-snowflake2Вручную: положить файлы
.github/skills/dbt-projects-on-snowflake/SKILL.md1Через Skills CLI (Vercel)
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a gemini-cli2Командой платформы
gemini skills install https://github.com/sfc-gh-dflippo/snowflake-dbt-demo --path .claude/skills/dbt-projects-on-snowflakeВручную: положить файлы
.devin/skills/dbt-projects-on-snowflake/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Через Skills CLI (Vercel)
npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a opencodeClaude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о dbt-projects-on-snowflake
Как установить dbt-projects-on-snowflake в Claude Code?
Выполните npx skills add sfc-gh-dflippo/snowflake-dbt-demo --skill dbt-projects-on-snowflake -a claude-code. Скилл окажется по пути ~/.claude/skills/dbt-projects-on-snowflake/SKILL.md.
Что делает dbt-projects-on-snowflake?
Помогает агенту развернуть dbt-проект как объект DBT PROJECT в Snowflake, запускать его команды и смотреть результаты в event-таблицах. Внутри также готовые SQL-запросы для поиска ошибок и метрик производительности.
Что нужно настроить до запуска мониторинга?
Event-таблицу, внешний доступ и интеграцию с Git API — шаги расписаны в references/SETUP.md. Пока таблица не настроена, скриптам из scripts/ нечего читать.
Исходный текст
SKILL.md · Apache-2.0
---
name: dbt-projects-on-snowflake
description:
Deploying, managing, executing, and monitoring dbt projects natively within Snowflake using dbt
PROJECT objects and event tables. Use this skill when you want to set up dbt development
workspaces, deploy projects to Snowflake, schedule automated runs, monitor execution with event
tables, or enable team collaboration directly in Snowflake.
---
# dbt Projects on Snowflake
Deploy, manage, and monitor dbt projects natively within Snowflake using web-based workspaces,
schema-level DBT PROJECT objects, and comprehensive event table telemetry.
## Quick Start
**Three Ways to Use dbt Projects:**
1. **Snowsight Workspaces** - Web-based IDE for interactive development
2. **DBT PROJECT Objects** - Deployed projects for production execution
3. **Snowflake CLI** - Command-line deployment and execution
## Setup
Complete setup instructions including prerequisites, external access integration, Git API
integration, and event table configuration are in `references/SETUP.md`.
## Deployment Methods
### Method 1: Snowflake CLI (Recommended)
```bash
# Deploy project
snow dbt deploy my_project --source .
# Execute commands
snow dbt execute my_project run
snow dbt execute my_project build
```
### Method 2: Snowsight
- Navigate to Projects → My Workspace
- Create new project from Git repository
- Configure profiles.yml
- Deploy as DBT PROJECT object
### Method 3: SQL Execution
Execute directly in SQL:
```sql
EXECUTE DBT PROJECT <db>.<schema>.<project> args='build';
EXECUTE DBT PROJECT <db>.<schema>.<project> args='build --full-refresh';
EXECUTE DBT PROJECT <db>.<schema>.<project> args='build --select tag:gold';
```
## Check and pin supported dbt versions
Run this to see the live list of versions currently available in the Snowflake managed runtime:
```sql
-- List the dbt versions supported by dbt Projects on Snowflake (Fusion + Core)
SELECT SYSTEM$SUPPORTED_DBT_VERSIONS();
```
`SYSTEM$SUPPORTED_DBT_VERSIONS()` returns each supported version along with its engine type. Use
this instead of relying on a hardcoded number — preview builds advance frequently.
The `DBT_VERSION` attribute implicitly selects the engine:
- **< 2.0** (e.g. `1.11.11`) → dbt Core (Python engine)
- **≥ 2.0** (e.g. `2.0.0-preview.186`) → dbt Fusion (Rust engine)
```sql
-- Set an account-wide default (a DBT_VERSION >= 2.0 selects the dbt Fusion engine)
ALTER ACCOUNT SET DEFAULT_DBT_VERSION = '2.0.0-preview.186';
-- Pin a project to a specific version at creation
CREATE OR REPLACE DBT PROJECT my_dbt_project
FROM '@my_stage/dbt_files'
DBT_VERSION = '2.0.0-preview.186';
-- Change an existing project's version
ALTER DBT PROJECT my_dbt_project SET DBT_VERSION = '2.0.0-preview.186';
```
Per-execution override (without changing the project definition):
```sql
EXECUTE DBT PROJECT my_db.my_schema.my_dbt_project
DBT_VERSION = '2.0.0-preview.186'
args='build';
```
Snowflake keeps versions supported even after dbt Labs deprecates them, so you are not forced into
immediate upgrades.
## Scheduling & Automation
For automated scheduling with Snowflake Tasks, see the "Optional: Schedule Automated Runs" section
in `references/SETUP.md`.
## Event Table Monitoring
### Event Table Configuration
Configure event tables following the Event Table Monitoring Configuration section in
`references/SETUP.md`. This enables OpenTelemetry-based monitoring of dbt project executions.
### Monitoring Queries
All monitoring scripts use parameterized event table references. Specify your event table location
when running:
```bash
# Example: Query recent executions
snow sql -f scripts/recent_executions.sql --enable-templating JINJA \
-D event_table=MY_DATABASE.MY_SCHEMA.EVENT_LOG
# Example: Check for errors
snow sql -f scripts/execution_errors.sql --enable-templating JINJA \
-D event_table=LOGS_DB.PUBLIC.DBT_EVENTS
# Example: Performance metrics
snow sql -f scripts/performance_metrics.sql --enable-templating JINJA \
-D event_table=MY_DATABASE.MY_SCHEMA.EVENT_LOG
```
**Core Monitoring:**
- **`recent_executions.sql`** - Lists recent dbt project executions with severity
- **`execution_errors.sql`** - Query ERROR logs to identify failures
- **`performance_metrics.sql`** - Query CPU and memory usage metrics
- **`trace_spans.sql`** - Query execution spans for timing analysis
- **`execution_summary.sql`** - Summarize executions by project with error counts
**Advanced Use Cases:**
- **`alert_failures.sql`** - Alert trigger for execution failures (returns error count)
- **`performance_regression.sql`** - Week-over-week performance comparison
- **`resource_usage.sql`** - CPU and memory consumption by project
- **`audit_trail.sql`** - Complete execution audit trail for compliance
### Event Table Structure
Event tables follow the [OpenTelemetry data model](https://opentelemetry.io/) with these key
columns:
| Column | Description |
| ----------------------- | --------------------------------------------------------------------------- |
| **TIMESTAMP** | UTC timestamp when event was created (end of time span for span events) |
| **START_TIMESTAMP** | For span events, the start of the time span |
| **TRACE** | Tracing context with `trace_id` and `span_id` |
| **RESOURCE_ATTRIBUTES** | Source identification: database, schema, user, warehouse, etc. |
| **SCOPE** | Event scopes (e.g., class names for logs) |
| **RECORD_TYPE** | Event type: `LOG`, `SPAN`, `SPAN_EVENT`, `EVENT`, `METRIC` |
| **RECORD** | JSON object with record-specific data (severity, metric type, span details) |
| **RECORD_ATTRIBUTES** | Event metadata set by Snowflake or code |
| **VALUE** | Actual log message, metric value, or null for spans |
### Best Practices
**Performance Optimization:**
- **Always filter by TIMESTAMP** to limit scanned data (reduces cost)
- **Use RESOURCE_ATTRIBUTES** for efficient filtering by project/database/schema
- **Archive old event table data** (>90 days) to separate tables
**Monitoring Strategy:**
1. Set event tables at **DATABASE level**, not account or schema
2. Configure appropriate log/trace/metric levels per schema
3. Always filter by `TIMESTAMP` to avoid scanning large event tables
4. Use `snow.executable.type = 'DBT_PROJECT'` to isolate dbt events
5. Leverage `RESOURCE_ATTRIBUTES` for filtering by project/database/schema
6. Monitor ERROR severity logs for immediate alerts
7. Use SPAN records to analyze execution timing and bottlenecks
**Alerting Priorities:**
- **High**: Any ERROR in execution, execution >2x historical avg, warehouse credit anomalies
- **Medium**: WARNING logs, test/model failures on critical models, performance trending down
- **Low**: INFO logs, scheduled job confirmations, performance metrics for analysis
### Event Table Troubleshooting
| Issue | Solution |
| ----------------------- | -------------------------------------------------------------- |
| No events captured | Verify event table set at DATABASE level with `ALTER DATABASE` |
| Too many events | Adjust `LOG_LEVEL`/`TRACE_LEVEL`/`METRIC_LEVEL` per schema |
| Slow monitoring queries | Always filter by TIMESTAMP first; consider archiving old data |
| Missing metrics | Set `METRIC_LEVEL = 'ALL'` for schema |
| Missing traces | Set `TRACE_LEVEL = 'ALWAYS'` for schema |
| Cannot see project name | Verify `snow.executable.type = 'DBT_PROJECT'` filter |
## Supported dbt Commands
| Command | Workspaces | EXECUTE DBT PROJECT | snow dbt execute |
| -------- | ------------------- | ------------------- | ---------------- |
| build | ✅ | ✅ | ✅ |
| run | ✅ | ✅ | ✅ |
| test | ✅ | ✅ | ✅ |
| compile | ✅ | ✅ | ✅ |
| seed | ✅ | ✅ | ✅ |
| snapshot | ✅ | ✅ | ✅ |
| deps | ✅ (workspace only) | ❌ | ❌ |
## Team Collaboration
**Flexibility:** Team members can use different development approaches simultaneously:
- Developer A: dbt Projects on Snowflake workspaces
- Developer B: dbt Cloud
- Developer C: Local VS Code with dbt CLI
- All check into the same Git repository
## Key Commands
| Command | Purpose |
| ------------------------------- | --------------------------- |
| `snow dbt deploy <name>` | Deploy project to Snowflake |
| `snow dbt execute <name> run` | Run dbt models |
| `snow dbt execute <name> build` | Run and test models |
| `snow dbt execute <name> test` | Run tests only |
| `snow dbt list` | List all dbt projects |
## Troubleshooting
For setup and deployment issues, see `references/SETUP.md`.
For monitoring issues, see the Troubleshooting table in the Event Table Monitoring section above.
## Related Skills
**Complementary Observability:**
- `dbt-artifacts` skill - For cross-platform execution logging and historical trend analysis
**When to use both together:**
- **dbt Projects on Snowflake** for real-time monitoring with OpenTelemetry event tables
- **dbt Artifacts** for cross-platform historical analysis and long-term metrics
**When to use one vs the other:**
- Use **dbt Projects on Snowflake** alone if you exclusively run dbt within Snowflake
- Use **dbt Artifacts** alone if you run dbt outside Snowflake (dbt Cloud, Airflow, local)
- Use **both** for comprehensive enterprise monitoring (real-time + historical)
---
## Resources
### Local Files
- **Monitoring Scripts**: `scripts/` - Ready-to-use parameterized SQL scripts for monitoring
- Core Monitoring: `recent_executions.sql`, `execution_errors.sql`, `performance_metrics.sql`,
`trace_spans.sql`, `execution_summary.sql`
- Advanced Monitoring: `alert_failures.sql`, `performance_regression.sql`, `resource_usage.sql`,
`audit_trail.sql`
- **Setup Guide**: `references/SETUP.md` - Complete step-by-step setup including event table
configuration and task scheduling
### Official Documentation
- dbt Projects Documentation:
<https://docs.snowflake.com/en/user-guide/data-engineering/dbt-projects-on-snowflake>
- Monitoring Guide:
<https://docs.snowflake.com/en/user-guide/data-engineering/dbt-projects-on-snowflake-monitoring-observability>
© sfc-gh-dflippo · Apache-2.0 · sfc-gh-dflippo/snowflake-dbt-demo