Content
76%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is well-organized and highly actionable with strong executable examples and clear sectioning, supported by well-signaled one-level-deep references. Its weaknesses are missing validation checkpoints in batch/security workflows and referenced bundle files that are absent from the bundle.
Suggestions
Add explicit validation/verification steps to the Practical Workflows — e.g. after RBAC grants run 'SHOW GRANTS TO ROLE analyst_role;' to confirm, and after MERGE upserts verify row counts — to lift workflow clarity above the destructive/batch cap of 3.
Consolidate the overlapping Proactive Triggers, Anti-Patterns, and Common Errors sections (which all restate the SELECT *, colon-prefix, deprecated-Cortex, and credential issues) into a single table or cross-linked section to reduce redundancy and save tokens.
Either ship the referenced files under references/ and scripts/ (snowflake_sql_and_pipelines.md, cortex_ai_and_agents.md, troubleshooting.md, snowflake_query_helper.py) or remove the references and inline the essential content, since the current bundle points to paths that do not exist.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely lean, using tables and code blocks to deliver Snowflake-specific rules (colon prefix, deprecated Cortex names, TARGET_LAG rules) rather than restating basics, but three sections — Proactive Triggers, Anti-Patterns, and Common Errors — rehash the same issues from slightly different angles and could be consolidated. | 4 / 5 |
Actionability | Provides copy-paste-ready, executable artifacts throughout — CLI commands in Quick Start, complete MERGE/Dynamic Table/Stream-Task SQL, a Snowpark Session builder, dbt config blocks, and Cortex function calls covering the common cases. | 5 / 5 |
Workflow Clarity | The three Practical Workflows give clear numbered sequences, but operations that are batch- or security-sensitive (RBAC grants, MERGE upserts, pipeline creation) lack explicit validation/verification checkpoints, capping workflow clarity at 3 per the destructive/batch rule. | 3 / 5 |
Progressive Disclosure | Structure is good: a focused overview with clearly signaled one-level-deep references (inline '> See references/...' callouts plus a Reference Documentation table) to three reference docs and one script, but the referenced bundle paths (references/, scripts/) do not exist in the skill bundle, a notable navigation gap. | 4 / 5 |
Total | 16 / 20 Passed |