Content
63%Weight 40%Scale 1-5Reviews 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 and well-sectioned, with concrete commands, tools, and config examples throughout, but it is a monolithic configuration manual rather than an overview: extremely verbose, frequently redundant, and with all reference material inlined instead of split into bundle files. Workflow sections (store, correct, debug) are clear, with only minor gaps like the unspecified bank-deletion step.
Suggestions
Split the configuration Reference table and deep-dive sections (pages/customPages, manageBankConfig, observationScopes, H scheduling) into reference files under references/, keeping SKILL.md to an overview with clearly signaled one-level-deep links.
Deduplicate repeated material: autoUpdate is explained at full length in both the Install section and the Reference row, and the repo-carried-config prohibition appears twice within the Opt-in only section — state each once and cross-link.
Give the exact command or control-plane location for deleting a bank in 'Resetting a repo's memory', since 'Delete its bank on the server' is the one non-executable instruction in an otherwise copy-paste-ready skill.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~740-line body includes paragraph-length table cells and clear redundancy — autoUpdate is fully explained both in the Install section and the Reference row, the repo-carried-config prohibition is stated twice within a few lines, and refresh-scheduling rationale repeats across sections. Not 1 because it never explains concepts Claude already knows; all content is plugin-specific but noticeably padded and repetitive. | 2 / 5 |
Actionability | Fully executable throughout: copy-paste npx install/update/uninstall/stats commands, exact tool signatures (hindsight_ingest_document(title, content), hindsight_search_knowledge_pages(query)), complete JSON config examples with placeholders explained, exact log paths, env var names, and a structured correction-document format. | 5 / 5 |
Workflow Clarity | Intent-to-tool mapping for storing, a 3-part correction procedure with explicit verification before fixing, and log-driven debugging with named events (session_start, deepen_started) are clearly sequenced. Minor gaps — e.g. 'Resetting a repo's memory: Delete its bank on the server' gives no command or location — keep it below the explicit-checkpoint anchor. | 4 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), so a ~500-line configuration and API reference is inlined monolithically in SKILL.md. Section headers and internal anchor links give real structure, so above the minimal-structure anchor, but content that clearly belongs in separate reference files is inline and there are no external references at all. | 3 / 5 |
Total | 14 / 20 Passed |