Content
92%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.
An exemplary lean informational skill: every token delivers claude-mem-specific knowledge, the structure exactly matches the description's three promises, and concrete commands appear wherever an action exists. The only improvement area is actionability detail, such as showing what an observation looks like or how to inspect the stored data.
Suggestions
Add a one-line sample observation so Claude can concretely illustrate what the observer writes when answering user questions.
Include a command for inspecting stored data (e.g., querying the SQLite database or listing observations) to make the 'where data lives' section fully actionable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every sentence conveys claude-mem-specific facts (observer model, second-session injection timing, ~/.claude-mem contents, exact data-flow destinations) that Claude could not know otherwise; there is no padding or explanation of known concepts. Anchor 5's 'every token earns its place' fits; anchor 4 would require some trimmable over-explanation, which is absent. | 5 / 5 |
Actionability | Provides concrete copy-paste commands ('/learn-codebase', 'npx claude-mem uninstall', 'npx claude-mem telemetry disable', 'DO_NOT_TRACK=1') and concrete locations (SQLite database, vector index), but leaves minor gaps such as a sample observation or a command to inspect stored data. This lands between anchor 4 ('mostly executable guidance... minor gaps') and anchor 5, closer to 4. | 4 / 5 |
Workflow Clarity | A simple single-purpose informational skill under 50 lines whose sections ('What it does', 'When it kicks in', 'Where data lives') directly mirror the three questions the description promises, making the single action unambiguous; the simple-skill exception applies and no destructive or batch operations are involved. | 5 / 5 |
Progressive Disclosure | Under 50 lines with no need for external references and no bundle files present; the three well-organized sections are sufficient structure, satisfying the under-50-lines guideline for a top score. | 5 / 5 |
Total | 19 / 20 Passed |