Conciseness

Quotes: the body opens with "Interleave elicitation, workpiece settlement, supported construction and checks" and never explains a concept Claude already knows; there is no filler, no library background, and instructions like "An explicit stop opens no new topic" carry maximal rule per token. It matches the 5 anchor — lean, assumes competence, every token earns its place — rather than 4, where trimming would still be possible.

Actionability

Quotes: concrete guidance abounds — "Read `references/pn-construction.md` and `references/checks.md` before beginning", "Settle the operational account with `mutate_workpiece`", and the three named dispositions ("Changed / Already represented / Blocked") each with a defined action. It is not 5 because some steps stay abstract: "Orient" gives no concrete exit test beyond "enough purpose and context", and "mounted Petrinaut tools" are never named, leaving minor gaps an executable example would close.

Workflow Clarity

Quotes: a clear sequenced lifecycle ("Orient", "Elicit or revise", "Maintain the workpiece", "Construct", "Check and deliver") with explicit validation gates — "run `references/checks.md`" after each changed fragment, "Apply `references/checks.md` whenever construction is prepared or attempted", and "Deliver a net only when the mounted tool path has produced and checked one" — plus error-recovery loops (Blocked disposition asks "the smallest resolving question", absent tools mean "limit the result to the workpiece"). This matches the 5 anchor's explicit validation steps, feedback loops, and decision checklist.

Progressive Disclosure

Quotes: detail is well deferred to clearly signaled one-level-deep files ("Read `references/profile.md` before substantive questions", "read `references/experiment-configuration.md` as part of the same disposition"), all four of which exist in the bundle. It drops below 5 because "Read `templates/workpiece.md`" and "`templates/workpiece.md` supplies the process-specific recording shape" point to a file that does not exist in the bundle — no `templates/` directory is present — a navigation gap that goes beyond the minor organization lapses the 5 anchor tolerates.

Specificity

Quotes: "Elicit or revise an operational process model, maintain its recoverable workpiece, and construct a checked SDCPN" lists several concrete, domain-specific actions covering the full lifecycle. It falls short of the 5 anchor because "analysis" of the artifacts is named only in the trigger clause rather than as a stated capability, and the load-bearing terms "SDCPN" and "Petrinaut" are never expanded, leaving a minor coverage gap.

Completeness

Quotes: the what is fully stated ("Elicit or revise an operational process model, maintain its recoverable workpiece, and construct a checked SDCPN") and the when is explicit with concrete triggers ("Use for a process-modelling interview, Petri net, or analysis or revision of either artifact"). This matches the 5 anchor — both what and when clearly and explicitly answered — and exceeds the 4 anchor, whose "when" is only serviceable.

Trigger Term Quality

Quotes: "process-modelling interview, Petri net, or analysis or revision of either artifact" — "Petri net" and "process-modelling interview" are phrases a user would naturally say. It sits at 4 rather than 5 because common variations are missing (e.g. "Petri nets" plural, "process model", "workflow"), and "SDCPPN"/"Petrinaut" are jargon no user would type.

Distinctiveness Conflict Risk

Quotes: "checked SDCPN when Petrinaut capabilities are available" and "Petri net" carve out a clear niche with distinct triggers; nothing here would plausibly fire for a generic document- or code-handling skill. It cannot be below 5 because no meaningful overlap with other skill domains exists.