Content
57%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a comprehensive lead intelligence skill with strong structural organization and good progressive disclosure to related skills and agents. Its main weaknesses are moderate verbosity (anti-patterns, channel descriptions that Claude would intuit), pseudocode rather than executable examples, and missing explicit validation checkpoints between pipeline stages. The weighted scoring models and ranking formulas are genuinely useful additions.
Suggestions
Add explicit validation/approval checkpoints between stages (e.g., 'Present signal scores to user for review before proceeding to mutual ranking' and 'Get user approval on outreach drafts before creating Mail drafts')
Replace pseudocode with more executable examples — show actual MCP tool call signatures for Exa and X API rather than invented function names
Trim the Anti-Patterns and Channel Rules sections significantly — Claude already knows not to use 'fake familiarity' or 'visible merge fields'; focus on project-specific constraints only
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is fairly long (~250 lines) and includes some unnecessary explanation (e.g., detailed descriptions of what each channel is for, anti-patterns that Claude would already know). The signal scoring table and mutual ranking model are useful but the overall document could be tightened by ~30%. The math formulas and pipeline diagram earn their place, but sections like 'Anti-Patterns' and channel descriptions are somewhat verbose. | 2 / 3 |
Actionability | The skill provides pseudocode rather than fully executable code (e.g., the Python snippets reference `web_search_exa()` and `search_recent_tweets()` without real API signatures). The weighted scoring tables and ranking formulas are concrete and useful, but the outreach drafting section gives guidelines rather than executable templates. The output format example for mutual ranking is helpful but most stages lack concrete output examples. | 2 / 3 |
Workflow Clarity | The 5-stage pipeline is clearly sequenced with a visual diagram, and each stage is well-delineated. However, there are no explicit validation checkpoints between stages — no 'verify signal scores before proceeding to mutual ranking' or 'confirm enrichment data quality before drafting.' For a pipeline that involves outreach (a semi-destructive/irreversible action), the lack of explicit approval gates beyond the final 'do not send automatically' note is a gap. | 2 / 3 |
Progressive Disclosure | The skill references external agents in `agents/` subdirectory, related skills (`brand-voice`, `connections-optimizer`, `social-graph-ranker`), and clearly signals when to use them. Content is well-organized with clear headers and the main SKILL.md serves as an overview that delegates appropriately. No deeply nested references. | 3 / 3 |
Total | 9 / 12 Passed |