Content
22%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 a thin, incomplete scaffold: persona padding, truncated and empty sections, placeholder solutions, and no executable guidance or validation. It needs concrete implementation detail and a real workflow before it can be actioned.
Suggestions
Replace the persona narrative with concrete implementation guidance — code or config for short-term/long-term/entity stores and the retrieval-consolidation flow.
Complete the truncated 'core principles' list and fill the empty Anti-Pattern sections with what to avoid and why, or remove them.
Add a validation/verification checkpoint for the cross-user isolation and bounded-growth risks flagged in the Sharp Edges table (e.g. an isolation test and a lifecycle/eviction check), and rewrite the placeholder '// ...' solutions as actual steps.
Replace the generic 'When to Use' sentence with concrete trigger conditions tied to the description.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The opening persona narrative ('You're a memory systems specialist who has built AI assistants that remember users across months...') and the conceptual aside ('memory is not just storage—it's about retrieval, relevance, and context') are padding Claude does not need, matching the 'noticeably verbose; several unnecessary explanations or padded sections' anchor. | 2 / 5 |
Actionability | Patterns are vague one-liners ('Store and update facts about entities', 'Include relevant memories in prompts') with no code or commands, and the sharp-edges 'solutions' are placeholder comments ('// Implement memory lifecycle management') rather than executable guidance, fitting 'minimal concrete guidance; high-level hints but missing the specific steps.' | 2 / 5 |
Workflow Clarity | There is no coherent multi-step workflow — the only numbered list is truncated mid-sentence ('1. Memory types differ—short-term, lo') — and despite the sharp-edges table flagging destructive/critical risks (cross-user memory access) there are no validation checkpoints, matching the 'steps missing or incoherent; no sequence; no validation' anchor. | 1 / 5 |
Progressive Disclosure | The body is sectioned (Capabilities, Patterns, Anti-Patterns, Sharp Edges, Related Skills) but several sections are empty shells (Anti-Patterns has only ❌ headers with no content) and there are no references at all; it has some structure but is incompletely organized, matching the 'some structure but could be better organized' anchor. | 3 / 5 |
Total | 8 / 20 Passed |