Conciseness

The ~20-line body is lean and assumes competence: a two-sentence rationale, five terse numbered steps, and three rules, with every line adding non-obvious guidance and no padding or concept over-explanation, matching anchor 5.

Actionability

The steps name concrete tools and bank naming patterns (`recall`, `grok-bot::<bot-name>`, `grok-bot::shared`, `list_banks`, `list_mental_models`, `get_mental_model`) plus a worked query example, but exact call syntax is not shown and execution depends on external skills (`memory-setup`, `bot-handoff`), leaving minor gaps as at anchor 4.

Workflow Clarity

The five steps are clearly sequenced with explicit conditionals — when to query which banks, when to read the mental model, what to do if a bank does not exist, and how to react when memory changes the plan — giving clear branches and an error-recovery path; the operation is read-only so no validation checkpoint is required, matching anchor 5.

Progressive Disclosure

The skill is under 50 lines with no bundle files and no content that warrants splitting; the body is organized into a rationale, Steps, and Rules sections with cross-references to related skills clearly signaled, which satisfies the simple-skill exception for a top score.

Specificity

The description names the domain and one concrete action — "Load relevant long-term memory from Hindsight before starting work" — but does not enumerate multiple distinct capabilities (e.g., recall across banks, mental model retrieval), leaving coverage incomplete as at anchor 3.

Completeness

It explicitly answers both what ("Load relevant long-term memory from Hindsight before starting work") and when ("Use at the start of every task or request, before planning or answering, and whenever the user mentions people, projects, decisions, preferences or anything done before") with concrete trigger phrases, matching anchor 5.

Trigger Term Quality

Natural trigger phrases are present ("people, projects, decisions, preferences", "anything done before", "at the start of every task or request"), but common synonyms users would say such as "remember", "recall", or "previously" are missing, matching anchor 4 rather than 5.

Distinctiveness Conflict Risk

The named system (Hindsight) and specific recall workflow give it a fairly distinct niche, but generic terms like "memory" and "anything done before" carry minor overlap risk with other memory-related skills, matching anchor 4.