Content
51%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 maximally token-efficient but is a hollow stub: its single instruction redirects to an absolute path outside the skill bundle that does not exist in this environment, leaving the skill with no actionable guidance, workflow, or bundled content of its own. As shipped, the skill cannot perform its stated function.
Suggestions
Inline the core guidance in SKILL.md (e.g., verify reviewer claims against the actual codebase, test suggestions before implementing, push back with evidence when feedback is technically wrong) rather than delegating everything to an external file.
Replace the machine-specific absolute path `~/.agents/skills/receiving-code-review/SKILL.md` with a bundle-relative reference (e.g., `references/process.md`) and include that file so the skill is self-contained and portable.
Add a short sequenced workflow with an explicit validation checkpoint (e.g., re-run tests after implementing each accepted suggestion) so the multi-step judgment process the description promises is actually present.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is a single lean instruction line ("Load and follow the shared skill at `~/.agents/skills/receiving-code-review/SKILL.md`.") with no padding and no explanations of concepts Claude already knows; the one-line HTML generator comment is de minimis. Not 4 because there is no over-explanation to trim — every instructional token earns its place. | 5 / 5 |
Actionability | The only guidance is a redirect to an absolute home-directory path that does not exist in this environment and is not part of the skill bundle, so the instruction cannot be executed as bundled and all task-specific steps are absent. Not 1 because the directive itself is concrete (a specific file path), not vague; not 3 because "some concrete guidance but incomplete" overstates a pointer whose target is unreachable and which conveys zero guidance on the actual task. | 2 / 5 |
Workflow Clarity | The only sequence is an implicit "load then follow" — every step of the actual task (evaluate feedback, verify claims, decide, implement) is delegated to a missing external file, with no validation checkpoints anywhere. Not 1 because the body is coherent and contains one clear directive; not 3 because there are no listed steps for the skill's actual multi-step judgment task, only a hollow redirect. | 2 / 5 |
Progressive Disclosure | The bundle contains no references/, scripts/, or assets/ directories, and the single reference ("~/.agents/skills/receiving-code-review/SKILL.md") is clearly signaled and one level deep but points outside the bundle to a machine-specific absolute path; the SKILL.md itself contains no overview content. Not 4 because "most content is appropriately placed" cannot hold when the bundle holds no content at all; not 2 because the reference is not buried and nothing is improperly inlined. | 3 / 5 |
Total | 12 / 20 Passed |