Conciseness

The body is dense and lean: every line carries tool-specific facts Claude cannot know (JSON envelope shapes, flag semantics, error codes like 'managed_resource_conflict', session-binding rules). It assumes Claude's competence and contains no padding or explanations of known concepts, matching the anchor 'Lean and efficient; every token earns its place'; level 4 would require minor instances of over-explanation, which are absent.

Actionability

Commands are copy-paste ready with real arguments and payloads ('opencli browser open https://example.com --session research', chrome call examples with full JSON argument strings), and executable discovery paths are provided ('opencli list <query>', 'opencli browser <command> --help -f json'). Examples cover the common cases (open/state/click/close, network variants, native chrome calls), matching the level-5 anchor rather than level 4's 'minor gaps'.

Workflow Clarity

The session lifecycle and interaction sequence (open --session, state, find/interact by ref, close) is clearly laid out, and validation checkpoints are present: 'When error.outcome is unknown, inspect the affected state before retrying a write', stale-ref recovery ('read fresh state or locate the element again'), and 'Check analysisComplete, script errors, omitted scripts and networkSource'. It is not level 5 because there is no explicit numbered workflow or end-of-task checklist — sequence is implied by section order — though it stays above level 3 because error-recovery feedback loops are explicitly stated, so the destructive/batch cap does not apply.

Progressive Disclosure

Sections are well organized and full API detail is delegated one level deep to clearly signaled discovery surfaces ('opencli list <query>', '--help -f json', 'chrome describe' listing signatures and documentation). It is not level 5 because the ~85-line body inlines flag-level reference detail (the paragraph of state/find/fill/wait flag semantics) with no external reference files organizing it, and the under-50-line simple-skill exception does not apply; it exceeds level 3 because the structure is consistently clear and navigation is easy.

Specificity

The description lists five concrete actions — 'Drive Chrome', 'inspect pages', 'interact with logged-in sites', 'capture API evidence', 'manage native tabs, groups and windows' — comprehensively covering the tool's page-driving, evidence-capture, and native-management surface. This matches the anchor 'Lists multiple specific concrete actions; comprehensive coverage' rather than the level-4 anchor, which requires minor coverage gaps.

Completeness

The 'what' is clear and concrete, but there is no 'Use when...' clause or equivalent explicit trigger guidance — the trailing sentence 'For reusable adapters, see opencli-adapter-author' is a cross-reference, not a usage trigger. Per the judging guidelines, a missing 'Use when...' clause caps completeness at 3; the description is not level 2 because the 'what' is neither vague nor absent.

Trigger Term Quality

Good natural keyword coverage ('Chrome', 'pages', 'logged-in sites', 'tabs', 'groups and windows') that users would plausibly say, but common variations such as 'browser', 'navigate', 'click', 'screenshot', or 'scrape' are absent. This fits the anchor 'Good keyword coverage; a few natural terms missing' and falls short of level 5's 'comprehensive coverage of natural terms including synonyms'.

Distinctiveness Conflict Risk

It names a specific tool and niche ('Drive Chrome with OpenCLI', 'native tabs, groups and windows') and explicitly disambiguates from the sibling skill via 'For reusable adapters, see opencli-adapter-author'. This is a clear niche with distinct triggers and minimal conflict risk, matching the level-5 anchor; it exceeds level 4, which allows minor overlap risk with closely related skills.