Conciseness

The body is lean and dense: path and permission tables, exact commands ('Check with `ls /volume` or `ls ~/"Cua Volume"`'), and only product-specific semantics Claude could not know (sync states, conflict naming, secret scanning). It assumes Claude's competence throughout with zero padding, matching anchor 5 exactly.

Actionability

Guidance is concrete and executable: exact mount paths, fallback commands, a full tool table with per-tool use ('volume_request_access | Ask the user for more (a folder, `r` or `rw`, and why)'), and precise behavioral rules ('Close files before telling the user they are ready'). It stops short of anchor 5 because no example invocation of the volume_* tools (e.g. a sample volume_write call or conflict filename) is shown.

Workflow Clarity

Sequences are clear with most checkpoints present: mount detection with explicit fallback (ls /volume, then ~/'Cua Volume', then tools), 'Before relying on one, check `volume_sync_status`', and 'Close files before telling the user they are ready'. Not 5, because the destructive tools (volume_delete, volume_restore) are listed without any verify-before-delete guidance, leaving a minor validation gap.

Progressive Disclosure

The body (~68 lines) has no bundle files and is organized into well-labeled sections (Where it is, Layout and what you may do, Tools, Sync and conflicts, Never put secrets) with everything appropriately inline at this size. It scores 4 rather than 5 because it slightly exceeds the under-50-line simple-skill threshold and keeps all sync/conflict detail inline rather than demonstrating any split, though nothing is buried or nested.

Specificity

The description lists several concrete actions with minor coverage gaps ('mounted as a folder (/volume on Linux, ~/Cua Volume on macOS)', 'your memory and outputs in your agent home, shared reference in public/', 'hand files to the user or to other agents'), which fits anchor 4 — but it uses second-person voice ('your memory and outputs in your agent home'), triggering the rubric's 1-point specificity penalty, landing it at 3. It is above anchor 3 because more than 1-2 concrete actions are named with platform-specific paths.

Completeness

Both 'what' ('Use Cua Volume, the one volume every Space and agent of this user shares... mounted as a folder... any program reads and writes it directly... outlive this Space... hand files to the user or to other agents') and 'when' ('Use when the user mentions the volume, Cua Volume, shared files, your home or memory, or saving results') are explicitly and concretely stated, matching the anchor-5 example pattern. Not 4, because the trigger guidance is fully explicit rather than vague.

Trigger Term Quality

The 'when' clause covers natural phrases users would say: 'the volume, Cua Volume, shared files, your home or memory, or saving results' — good keyword coverage with synonyms (volume, Cua Volume, shared files). It falls short of anchor 5 because common variations like 'sync', 'persist', or 'share with my other device/Mac' are missing.

Distinctiveness Conflict Risk

The 'Cua Volume' niche is clearly distinct from generic skills, but broad trigger terms like 'shared files', 'memory', and 'saving results' create minor overlap risk with memory- or file-saving-related skills, matching anchor 4. Not 5, because those generic terms could pull in the wrong skill; not 3, because the product-specific triggers dominate.