Content
63%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 delivers a well-sequenced, concrete, multi-platform workflow with genuinely executable examples, and the bundle references are real and well-organized. Its main weaknesses are redundancy — Step 5 reproduces two large data tables that the reference files already contain — and a missing local verification step plus an unspecified staging-smoke assertion mechanism.
Suggestions
Remove the per-runtime cold-start and integration-timeout tables from Step 5 and link directly to references/cold-start-budget.md and references/timeout-budgets.md instead; keep only one or two illustrative rows (e.g., the API Gateway 29s ceiling) inline.
Add an explicit validation step after Step 4 (e.g., run `npx jest tests/integration/` locally and fix failures before writing the CI config and matrix), and show a short staging-smoke test example that asserts the Step 5 budgets (cold_start_p99_ms / warm_p99_ms) against deployed functions.
Link references/timeout-budgets.md directly from SKILL.md's References section rather than only via cold-start-budget.md, keeping the reference graph one level deep; include a Python/jest-setup counterpart for the `conftest.py` entry in the scaffold layout.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean and table-driven (yq one-liners, command mappings, anti-pattern tables), but Step 5 says "derive a budget from references/cold-start-budget.md" and then inlines ~35 lines of data that duplicate the reference files almost verbatim — the per-runtime cold-start table ("AWS Lambda Java 11 (512MB, no SnapStart) | 1.5-6s") repeats cold-start-budget.md's table, and the integration timeout table ("API Gateway ... 29s hard") repeats timeout-budgets.md's cascade. This is more than the minor trimmable instances of the score-4 anchor, fitting the score-3 anchor for content that could be tightened. | 3 / 5 |
Actionability | Mostly executable and copy-paste ready: the full jest test file (import, fixtures, timeout-budget assertion), yq inventory commands, `sam local generate-event apigateway aws-proxy` event commands, directory layout, and a complete GitHub Actions workflow. Not 5 because of minor gaps: no Python test example despite `conftest.py` appearing in the layout, and the Step 5 budgets yaml "become assertion thresholds in staging-smoke tests" with no scaffold or code showing how (the staging-smoke job runs `npx jest tests/staging-smoke/` against a directory the skill never populates). | 4 / 5 |
Workflow Clarity | A clearly sequenced 7-step process (inventory IaC → pick emulator → generate events → scaffold tests → set budgets → CI → coverage matrix) with most checkpoints present: the two-tier CI ("local-fast tests on every PR; deploy + smoke on main") and the Step 7 coverage matrix with a Gaps section act as verification. Not 5 because there is no explicit local validation loop (e.g., run the generated tests and fix failures before emitting the matrix) — verification is only embedded in the emitted CI config, leaving a minor gap between generation and confirmation. | 4 / 5 |
Progressive Disclosure | References are clearly signaled (Step 5 link and a References section) and both referenced files exist, but two structural problems remain: content that already lives in the reference files is duplicated inline in Step 5 (the cold-start and integration-timeout tables), and references reach two levels deep — SKILL.md → references/cold-start-budget.md → references/timeout-budgets.md, which SKILL.md itself points to only indirectly ("timeout budgets in its references/timeout-budgets.md") rather than linking directly. This fits the score-3 anchor: structure present and references signaled, but content that should stay separate is inlined and organization could improve. | 3 / 5 |
Total | 14 / 20 Passed |