Content
62%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-structured, highly actionable skill with excellent workflow clarity and validation checkpoints. Its main weakness is verbosity — it spends significant tokens on motivational framing, explanations of why each step matters, and repeated concepts that Claude doesn't need spelled out. The content would benefit from aggressive trimming (likely 40-50% reduction possible) while preserving all the concrete guidance, which is genuinely strong.
Suggestions
Cut all motivational/justification text (e.g., 'A bad problem frame is the most expensive mistake...', 'This matters because...', 'Write it like it matters') — Claude doesn't need persuading, just instructing.
Deduplicate repeated concepts: falsifiability is covered in Step 3e, Step 3.5, Step 4, and Common Issues — consolidate into one authoritative location.
Extract the output template and the structured critique framework (Step 3.5) into separate referenced files to reduce the main skill's token footprint and improve progressive disclosure.
Remove explanatory asides like 'Don't fire five clarifying questions — pick the most important one and ask it well' — state the rule directly: 'Ask at most ONE clarifying question.'
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~250+ lines. It includes extensive motivational framing ('A bad problem frame is the most expensive mistake...'), explains reasoning Claude already understands ('This matters because a problem framing that ignores...'), and repeats concepts across steps (e.g., falsifiability is mentioned in Step 3e, Step 3.5, Step 4, and Common Issues). Much of the philosophical justification and tone-setting could be cut without losing actionability. | 1 / 3 |
Actionability | The skill provides highly concrete, specific guidance at every step: exact file paths to read, specific Headout funnel terminology (TOFU/BOFU, S2O, C2O, GBV), a complete output template with field-level instructions, a worked example showing input→reframe→hypothesis, and specific anti-patterns with fixes. A PM or Claude could follow this and produce a well-structured problem frame. | 3 / 3 |
Workflow Clarity | The workflow is clearly sequenced (Steps 1→2→2.5→3→3.5→4→Output) with explicit validation checkpoints: Step 2.5 has a completion criterion ('no answer would materially change the problem direction'), Step 3.5 is a structured critique with gap/impact/recommendation format, and Step 4 is a deliberation checkpoint before output. The feedback loop in Step 3.5 ('Each gap must be resolved or explicitly acknowledged') is well-defined. | 3 / 3 |
Progressive Disclosure | The skill references external context files appropriately (CLAUDE.md, company.md, historical-pipeline.md, active-pipeline.md), but the body itself is monolithic — all content is inline in one long file with no references to supporting documents for detailed guidance. The output template, common issues, and critique framework could be split into separate files to reduce the main skill's token footprint. | 2 / 3 |
Total | 9 / 12 Passed |