Content
100%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.
An exemplary instruction-only skill body: two clearly sequenced workflows with planning, validation, and error-recovery feedback loops, exact tool and command names throughout, and detail correctly pushed into two well-signaled one-level-deep reference files. No padding, no concepts Claude already knows, and staleness handled by deferring to the live schema.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~55-line body is lean and assumes Claude's competence: every sentence carries a tool name, flag, or behavioral rule ('Installs can outlast tool timeouts, so create with install: false'), with no explanations of concepts Claude already knows and tables used where prose would bloat. | 5 / 5 |
Actionability | Guidance is fully executable: exact MCP tool names (bts_plan_project, bts_get_schema, bts_add_addons), exact commands (npx create-better-t-stack@latest schema --name createInput, create-json with dryRun), and exact flags (install: false), with complete copy-paste payloads one reference away in references/cli.md. | 5 / 5 |
Workflow Clarity | Both the 7-step create and 4-step change workflows have explicit validation checkpoints and feedback loops: plan/dryRun before writing, 'If the plan fails, its error names the conflict. Fix the config and plan again', user confirmation ('get a yes'), and 'Prove it runs... If a step fails, report the exact error.' | 5 / 5 |
Progressive Disclosure | The body is a well-organized overview that splits detail into two clearly-signaled, verified, one-level-deep references — references/options.md (values and compatibility rules, with the schema named as source of truth) and references/cli.md (full payloads, flag form, Stack Builder keys) — with no further nesting. | 5 / 5 |
Total | 20 / 20 Passed |