Content
42%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill provides highly actionable, specific guidance with named MCP tools and concrete parameters, which is its strongest quality. However, it is excessively verbose — explaining obvious behaviors, including lengthy conversational scripts, and inlining large output templates that bloat the token budget. The lack of any supporting bundle files means everything is crammed into one long document with no progressive disclosure, and the workflow lacks validation/error-handling checkpoints for a multi-step process involving parallel API calls.
Suggestions
Cut the content by at least 50% — remove explanations of what MCP tools do, the conversational prompt template in Step 2b, and the detailed descriptions of how to search each platform (Claude knows how to use tools it has access to).
Extract the output template (relationship summary format) into a separate TEMPLATE.md file and reference it from the main skill, and move the cross-platform enrichment details into a CROSS_PLATFORM.md reference file.
Add validation checkpoints: what to do if MCP calls return no results, how to handle contacts not found, and how to verify data quality before presenting the summary.
Condense the cross-platform section to a brief list of tool categories to check rather than spelling out search strategies for each platform individually.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~200+ lines. It over-explains concepts Claude already understands (how to detect MCP tools, how to interpret read receipts, what a CRM is). The cross-platform section (Step 2b) alone is massive and largely describes obvious behavior. The conversational prompt template ('I've pulled your email and calendar history...') and extensive output template add significant token bloat. Much of this could be condensed to a fraction of its size. | 1 / 3 |
Actionability | The skill names specific MCP tool calls (Superhuman_Mail.list_threads, get_read_statuses, get_thread, query_email_and_calendar, create_or_update_draft, get_availability, create_or_update_event) with concrete parameters and filters. The output template is detailed and copy-paste ready, and the workflow specifies exact numbers (50 threads, 5-10 for read receipts, 90 days default). | 3 / 3 |
Workflow Clarity | The steps are clearly sequenced (identify target → gather data → cross-platform enrichment → build summary → offer actions), but there are no validation checkpoints. There's no guidance on what to do if MCP calls fail, return empty results, or if the contact can't be found. For a multi-step process involving parallel API calls and potentially destructive actions (drafting emails), the absence of error handling and verification steps is a gap. | 2 / 3 |
Progressive Disclosure | Everything is in a single monolithic file with no references to supporting documents. The cross-platform context section, the output template, the multi-contact view, and the guidelines could all be split into separate referenced files. The inline output template alone is ~30 lines that could be a separate TEMPLATE.md. No bundle files exist to offload content to. | 1 / 3 |
Total | 7 / 12 Passed |