Content
67%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.
This is a highly actionable, defensively thorough skill: concrete commands, explicit validation checkpoints, bounded recovery, and a well-structured one-level reference bundle. Its main weakness is token efficiency — the body is roughly twice as long as it needs to be due to policy restated across sections and dense multi-clause prose.
Suggestions
Consolidate the depth-selection policy, which is currently stated in four places (the 'Choose when and how deeply to review' section, 'Checkpoints and final handoff', 'Configuration', and 'Guardrails'), into one canonical section referenced from the others.
Move the runtime compatibility gate and 1.x→2.x migration detail (currently ~20 dense lines) into a reference file, keeping a short gate summary plus link in the body.
Break the long multi-clause policy sentences into short imperative rules so each rule can be read once instead of re-parsed.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body avoids explaining concepts Claude already knows and carries high information density per line, but it runs ~490 lines with the depth-selection policy restated in four places (the selection table, checkpoints, Configuration, and Guardrails) and long multi-clause sentences ('Only when the runtime needs an update, before any update check/change, test whether QODO_UPDATE_BASE_URL is set; never print its value') that require re-reading. Not 2 because the problem is repetition and density, not fluff or padding. | 3 / 5 |
Actionability | Provides concrete, executable artifacts: a version probe, a provenance-stamped `qodo read whoami` call, a fully guarded async-submission script with jq checks, a trap-based polling loop with an exit-code table, and a session-context JSON example. Not 5 because the examples are explicitly labeled 'illustrative' with a 'Confirm the exact flags with `qodo review --help`' caveat, so they are not guaranteed copy-paste ready. | 4 / 5 |
Workflow Clarity | The lifecycle is well sequenced with explicit validation gates (runtime version probe, auth check, merge-base SHA resolution, digest/byte-count verification, exit-code handling) and genuine feedback loops (fix → verify → re-review, bounded recovery with retry limits). Not 5 because the ordering scatters the flow — 'Choose the execution path' and 'Choose execution for the selected review' overlap, and Quick start appears mid-document after the runtime gate. | 4 / 5 |
Progressive Disclosure | All eight referenced files (skill-updates, windows-shells, windows-launcher, review-turn-context, connected-progress, finding-results, context-delivery, local-triage) exist, are one level deep, and are well-signaled with descriptive link text at the point of need; the `#cli-examples` anchor is real. Not 5 because the body inlines substantial dense policy — the runtime compatibility gate, 1.x→2.x migration detail, and async cost notes — that belongs in reference files, making the overview itself long. | 4 / 5 |
Total | 15 / 20 Passed |