Content
38%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 reads like product documentation rather than an executable skill: buzzword-heavy overview sections, pseudocode dependent on undefined helpers, workflows without validation checkpoints, and a dangling reference to a nonexistent playbook file. It does offer a recognizable structure and some concrete commands, but little of it is directly actionable.
Suggestions
Delete the Role Statement, Context Overview, Future Roadmap, and Integration Patterns sections — they are marketing copy that instructs nothing — and cut the bullet lists of technique names to the ones actually used in workflows.
Replace the pseudocode with runnable guidance: either define or remove the undefined helpers (VectorDatabase, prioritize_components, estimate_tokens, etc.), and document how the `context-restore` CLI is obtained/invoked before showing bash examples.
Fix the dangling reference: create resources/implementation-playbook.md and move the 8 advanced strategy sections and code samples into it, leaving SKILL.md as a concise overview with well-signaled one-level-deep links.
Add explicit validation checkpoints to both workflows (e.g., how to verify restored context against the current codebase, and what to do on mismatch) so steps are verifiable rather than abstract.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Several sections are pure padding: a "Role Statement" ("Expert Context Restoration Specialist focused on..."), a "Future Roadmap" with buzzwords ("Quantum-inspired vector search algorithms", "Self-healing context reconstruction"), and bullet lists that name techniques ("Cryptographic context signatures") without instructing anything. This fits anchor 2 (noticeably verbose, several padded sections); it avoids anchor 1 because it never dwells on basic concepts Claude already knows. | 2 / 5 |
Actionability | The code blocks are non-executable pseudocode — every function depends on undefined helpers (VectorDatabase, rank_and_filter_contexts, calculate_composite_score, prioritize_components, estimate_tokens, load_component) — and the bash examples invoke a `context-restore` CLI that is never set up or installed. Concrete guidance exists but is incomplete, matching anchor 3 exactly; it falls short of anchor 4's executable, minor-gaps bar. | 3 / 5 |
Workflow Clarity | Two numbered workflows are listed ("1. Retrieve most recent project context ... 4. Generate resumption summary") but the steps are abstract — "Validate context against current codebase" names no command or check, and there are no validation checkpoints, error-recovery loops, or feedback mechanisms. Steps present with validation gaps fits anchor 3; it is not 2 because a genuine, ordered sequence with some specificity does exist. | 3 / 5 |
Progressive Disclosure | The body inlines ~180 lines of advanced material (8 strategy sections, integration patterns, roadmap) that clearly belong in a separate reference file, and its sole external reference — "If detailed examples are required, open `resources/implementation-playbook.md`" — points to a file that does not exist (no resources/ or bundle directories are present). This matches anchor 2 (content that belongs in separate files is inlined); the dangling reference is worse than anchor 3's merely unclear signaling. | 2 / 5 |
Total | 10 / 20 Passed |