Content
75%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 lean, executable workflow with concrete commands, a clear phase sequence, and real validation checkpoints, plus a correctly wired one-level-deep bundle reference. Its only weaknesses are mild redundancy around the hunt-skill responsibility split and the duplicated commit-message content between the inline example and the template file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient — terse command blocks with purposeful comments and a concrete commit example that earns its tokens — though the Phase 2 "The hunt will / BUT the hunt does NOT" responsibility listing is somewhat repetitive, and the Note bullets restate the template's rules. | 4 / 5 |
Actionability | Every phase contains concrete, executable shell commands, including a fully filled-in commit message example, three example branch names, and a complete gh pr create heredoc; only minor gaps remain (PR title placeholder, {MODEL} substitution instruction). | 4 / 5 |
Workflow Clarity | Phases 0-4 are clearly sequenced with explicit checkpoints ("STOP if working directory is dirty", "CRITICAL: ... you MUST return here to Phase 3") and validation delegated to /lading-preflight and the review skill's pass/fail judging; it lacks a feedback loop for commit/push/PR failures, keeping it below 5. | 4 / 5 |
Progressive Disclosure | The single bundle file (assets/commit-template.txt) exists and is referenced once by full path — one level deep and clearly signaled — and the body is well-sectioned into phases; the inlined commit example largely duplicates the template file's content, a minor organization gap. | 4 / 5 |
Total | 16 / 20 Passed |