Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is highly actionable with strong workflow checkpoints and an exemplary progressive-disclosure structure that points to real sub-docs. Its main weakness is motivational padding that competes with context without adding operational value.
Suggestions
Trim the hortatory read-the-docs prose (e.g. the 'STOP', 'This is not optional', and all-caps 'JUST READ MORE' passages) to imperative one-liners; the routing table already carries the instruction.
Consider moving the long provider-playbook link list (Section 'Provider Playbooks') into a references file and linking it once, since SKILL.md is meant to be the routing layer rather than a provider index.
Consolidate the repeated read-the-sub-doc reminders in Sections 2 and 5 into a single canonical routing note to reduce redundancy.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly dense and operational (exact CLI flags, routing tables, approval templates) but padded with hortatory prose motivating the agent to read docs — e.g. "reading a doc for 5 seconds saves you from 10 failed tool calls", "This has happened repeatedly", and "READING MULTIPLE DOCS IS A GREAT IDEA AND OFTEN SUPER ESSENTIAL. JUST READ MORE." — which Claude does not need. | 2 / 3 |
Actionability | Provides fully executable, copy-paste-ready guidance: concrete commands (`deepline tools search`, `deepline enrich --rows 0:1`), exact flag syntax, a verbatim approval template, and billing commands with specific examples throughout. | 3 / 3 |
Workflow Clarity | Multi-step processes are clearly sequenced with explicit validation checkpoints and feedback loops for batch/paid operations — pilot → request approval → full run only after approval, "If pilot fails, fix and re-run until successful", and the over-provision-then-filter Do/Don't list. | 3 / 3 |
Progressive Disclosure | Clear Level 1/2/2.5/3 documentation hierarchy with a task→doc routing table, a recipes table, and a provider-playbook link list; references are one level deep, clearly signaled, and the referenced bundle paths (recipes/*.md, enriching-and-researching.md, provider-playbooks/*.md, references/, scripts/) are real and well organized. | 3 / 3 |
Total | 11 / 12 Passed |