Content
87%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a high-quality skill that delivers dense, opinionated, domain-specific guidance efficiently. Its greatest strengths are conciseness (no wasted tokens explaining what Honeycomb or MCP tools are), actionability (concrete expressions, field names, and heuristics), and excellent progressive disclosure to reference files. The main area for improvement is making the query-building workflow more explicitly sequenced with decision points and validation steps.
Suggestions
Consider adding an explicit numbered workflow for the common investigation pattern (e.g., 1. find_queries → 2. find_columns to validate fields → 3. broad COUNT/GROUP BY → 4. interpret results → 5. narrow with WHERE → 6. HEATMAP if P99/P50 ratio is high → 7. run_bubbleup for outliers) with decision points at each step.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is lean and efficient throughout. It assumes Claude's competence with Honeycomb and MCP tools, explicitly stating it won't re-document tool parameters. Every section delivers domain-specific insight (e.g., 'Never use AVG for latency') rather than explaining basic concepts. The guardrails in calculated fields are all non-obvious, earned knowledge. | 3 / 3 |
Actionability | Provides concrete, copy-paste-ready patterns: specific operation choices with exact field names (e.g., `P99(duration_ms)`), calculated field expressions (e.g., `MUL(IF($error, 1, 0), 100)`), specific tool calls (`find_columns`, `find_queries`, `run_bubbleup`), and precise heuristics (P99/P50 > 10x). The operation selection table maps questions directly to specific operations. | 3 / 3 |
Workflow Clarity | The 'Before Every Query' section provides pre-flight checks and the 'Start broad, narrow with WHERE' principle gives a general workflow, but there's no explicit sequenced workflow with validation checkpoints or feedback loops. For a query-building skill this is less critical than for destructive operations, but the investigation flow (broad → narrow → interpret → bubbleup) could be more explicitly sequenced with decision points. | 2 / 3 |
Progressive Disclosure | Excellent structure: the main file provides an opinionated overview with key patterns, then clearly signals five reference files for deeper content (calculated fields, relational fields, visualize operations, query examples, result interpretation). References are one level deep, well-labeled with descriptions, and cross-references to related skills are included. Navigation is straightforward. | 3 / 3 |
Total | 11 / 12 Passed |