Content
96%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.
The body is a minimal, fully executable launcher: one command with defined substitutions and explicit handling for every outcome branch, with zero wasted tokens. Its only structural weakness is that the overview never surfaces the bundle's reference files, leaving navigation entirely to the render script's output.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is 13 lines with zero padding and no explanations of concepts Claude already knows — every line instructs ("Run the following command exactly once without changing the current working directory", "On any other output or failure, including `uv` being unavailable, report the command output and stop"). Matches the 5 anchor ('every token earns its place'). Not 4: there is nothing that could be trimmed. | 5 / 5 |
Actionability | A fully executable command is given verbatim — `uv run --no-cache "{project-root}/_bmad/scripts/render_skill.py" --project-root "{project-root}" --skill "{skill-root}"` — with explicit substitution rules ("Replace `{project-root}` with the nearest folder containing `_bmad/`, from the working directory upward") and branches covering the common failure cases (script missing, HALT, uv unavailable). Not 4: the only placeholders are explicitly defined, leaving no gaps between this and copy-paste-ready. | 5 / 5 |
Workflow Clarity | A single unambiguous action with exhaustive outcome handling: success ("The command prints the skill's workflow: follow it"), halt ("`HALT: <reason>`: report the reason and stop"), missing script ("Offer to set it up using `bmad` skill, then run the command again"), and other failures ("report the command output and stop. Do not run any workflow source directly"). Per the simple-skill note, a single unambiguous action with explicit branch directives merits 5; the operation is non-destructive and non-batch, so the validation cap does not apply. Not 4: every outcome branch has an explicit directive — no checkpoints are left implicit. | 5 / 5 |
Progressive Disclosure | Scored against the actual bundle: the content is appropriately split one level deep (5 phase docs in references/ — acceptance-verdict, aggregate-views, evidence-gathering, retro-document, team-discussion — plus scripts/git_evidence.py and its tests), stitched by the render script via one-level `{{ rendered(...) }}` inlines rather than nested link chains. However, the body itself signals none of these files — a reader of SKILL.md alone cannot see the skill's shape without running the renderer. Not 5: the references are not 'well-signaled' from the overview; not 3: nothing is inlined that belongs in a separate file, and navigation via the rendered workflow is one level deep. | 4 / 5 |
Total | 19 / 20 Passed |