rill-model

Detailed instructions and examples for developing model resources in Rill

1.32x

Quality

43%

Does it follow best practices?

Impact

94%

1.32x

Average score across 3 eval scenarios

Securityby

Advisory

Suggest reviewing before use

Optimize this skill with Tessl

npx tessl skill review --optimize ./skills/rill-model/SKILL.md

Evaluation results

100%

25%

Ad Impressions Pipeline: ClickHouse Incremental Model

ClickHouse incremental S3 model with LowCardinality, TTL, indexes, and dev partition

Criteria

Without context

With context

output.order_by present

100%

S3 credentials via .env templating

100%

LowCardinality for categorical strings

100%

TTL for data retention

100%

Performance index defined

100%

Partition-based incremental

70%

100%

Not append strategy

100%

output.partition_by set

100%

materialize: true

100%

Dev partition configured

100%

Dev partition uses time range not row limit

100%

84%

22%

Multi-Format Log Ingestion Pipeline

Multi-format DuckDB pipeline with materialization, CSV/JSON options, and ref() function

Criteria

Without context

With context

JSON: auto_detect and format='auto'

40%

100%

CSV: multiple read options

100%

Source models materialized

100%

Enriched model materialized

100%

Recent view NOT materialized

100%

ref() used in derived models

100%

No trailing semicolons in SQL

100%

Dev partition on source models

No dev partition on derived model

100%

Dev partition by time range

100%

type: model declared

50%

100%

22%

Prototype Sales Dashboard Model with Synthetic Data

Synthetic data prototype model with correct row count, time range, and distributions

Criteria

Without context

With context

Timestamp column included

100%

6-12 months of data

100%

Approximately 10,000 rows

100%

Realistic timestamp spacing

86%

100%

Diverse categorical values

100%

No trailing semicolon in SQL

100%

Realistic numeric distributions

100%

type: model declared

50%

100%

Repository: rilldata/agent-skills
Commit: 65ccd1f

Evaluated: 7 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Ad Impressions Pipeline: ClickHouse Incremental Model Multi-Format Log Ingestion Pipeline Prototype Sales Dashboard Model with Synthetic Data

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.