Content
73%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.
Operationally excellent: concrete, executable commands on both MCP and CLI paths, tightly validated workflows with dry-run/read-back feedback loops, and correctly structured one-level-deep references. The main liability is token efficiency — heavy repetition of the same approval, idempotency, and refresh gates across many sections makes the body noticeably verbose.
Suggestions
Consolidate the approval/dry-run/idempotency rules, currently restated in Instructions, Preflight, Guardrails, 'Build one reviewable proposal', and 'Preview and apply', into a single 'Write gates' section referenced once from each flow.
Move the runtime compatibility gate's upgrade-recovery detail (QODO_UPDATE_BASE_URL checks, latest-2 route, checksum-pinned installer fallback) into a reference file like the Windows guides, keeping only the unadorned version probe and minimum-version statement in SKILL.md.
Trim repeated caveats ('do not switch to a different CLI identity or tenant', 'refresh once and recheck') to a single statement each, and tighten long multi-clause sentences such as the sandbox auth diagnostic paragraph.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | At ~475 lines the body restates the same operational gates across sections — approval-reuse rules appear in Instructions, Preflight, Guardrails, 'Build one reviewable proposal', and 'Preview and apply'; catalog-refresh-once and never-switch-identity rules each repeat three or more times — and the runtime compatibility gate is ~25 lines of tangled conditional prose. This is 'noticeably verbose; several unnecessary explanations or padded sections' rather than the mostly-efficient 3 anchor, and time-sensitive version pins ('2.0.0 or newer', '--skill-version 1.0.9') sit outside any deprecated section. | 2 / 5 |
Actionability | The skill provides copy-paste-ready commands for both paths — 'qodo rules create --name "..." --category "..." --severity warning ... --scopes "/owner/repo/" --json', 'qodo review-skills bulk --args '{"operation":"set_scope",...,"dry_run":true}'' — plus the exact fallback invocation '"${QODO_HOME:-$HOME/.qodo}/bin/qodo"', specific error codes ('MT-RATE-LIMITED'), and operation tables. Examples cover the common cases across create, edit, scope, triage, discovery, and configuration, matching the fully-executable 5 anchor. | 5 / 5 |
Workflow Clarity | Multi-step flows are explicitly sequenced (preflight → resolve target → dry-run preview → confirm → mutate once → read back and report) with validation checkpoints throughout: dry-run matched-count inspection before every bulk write, read-back comparison of every affected id, and a bounded retry loop for uncertain outcomes ('read back the target before retrying... Retry at most once'). The destructive/batch cap does not apply because irreversible operations like 'reject' require dry-run, a matched-count display, and a summary of what will be permanently lost. | 5 / 5 |
Progressive Disclosure | All three bundle files (references/skill-updates.md, references/windows-shells.md, references/windows-launcher.md) are real, well-signaled, conditionally loaded ('Read only when...'), and exactly one level deep. It misses the 5 anchor because the SKILL.md body itself inlines substantial flows — the runtime upgrade recovery detail and the multi-page source-repository discovery procedure — that would belong in reference files like the Windows guides. | 4 / 5 |
Total | 16 / 20 Passed |