Conciseness

Dense, table-driven, and free of padding — every section delivers Dagster-specific semantics Claude would not reliably know (is_required=False, no trigger_rule, condition syntax). Falls short of 5 because the trigger-time-vs-runtime rule is stated three times (Overview, mapping table, Gotchas) and 'DynamicOut is fan-out, not branching' is likewise repeated in three places.

Actionability

Fully executable paired examples for both main patterns (Dagster Python + Orchestra YAML), a concrete variables table with real expression syntax ('task_groups['id'].all().status == 'SUCCEEDED'', 'ORCHESTRA.PIPELINE_RUN_TASKS['decide'].OUTPUTS['has_data']'), a functions list, and a copy-paste-ready quick-reference card covering the common cases.

Workflow Clarity

The decision procedure is stated crisply up front ('is the branch decided at trigger time (use inputs:) or at runtime (use an upstream task's OUTPUTS)?') and patterns are organized around that dichotomy, with gotchas covering edge cases like SKIPPED propagation and ANDed group/task conditions. Not a 5 because there is no explicit step-by-step conversion checklist or verification step, though this is a mapping skill rather than a destructive workflow.

Progressive Disclosure

A single well-sectioned file with clear headers, no nested references, and external doc links clearly listed in a References section; all inline content is core to the task. Not a 5 because the condition-expression reference (variables, status values, built-in functions) is reference material that could live in a bundle file, though at this size keeping it inline is defensible.

Specificity

Enumerates several concrete trigger patterns ('ops with multiple Out/branching outputs that are conditionally yielded', '@op functions that decide which downstream op runs', 'DynamicOut fan-out', 'conditional asset materialization (AutomationCondition / skip logic)'), matching the several-specific-actions anchor. Not a 5 because the action the skill performs is never named, and not a 3 because it lists more than 1-2 concrete items.

Completeness

The 'when' is explicit and concrete ('Use this skill when...', 'Triggers: any Dagster op that conditionally yields one of several Outs'), but the 'what' — what the skill actually does (converts these to Orchestra condition expressions) — is never stated and is only inferable from the skill name. Per the guideline to score only what is explicit, the 'what' is weakly implied rather than present.

Trigger Term Quality

Includes natural terms a user with this migration problem would say: 'Dagster', 'conditional execution', 'branching', 'Out', 'DynamicOut', 'skip logic', 'asset whose execution is gated on a condition'. Good coverage but missing common variations like 'conditional output' or 'if/else op', so it sits between the 3 and 5 anchors, closer to 4.

Distinctiveness Conflict Risk

'any Dagster op that conditionally yields one of several Outs' carves a clear niche, but the description also claims DynamicOut fan-out and AutomationCondition as triggers, which the body itself routes to sibling skills (matrix, sensors), creating minor overlap risk with closely related skills.