Content
77%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 an unusually strong operational skill: executable commands, a fail-closed readiness checklist, and explicit feedback loops for nearly every failure mode. Its weaknesses are structural — a monolithic ~290-line file with no reference split, repeated command prefixes, and inline version-specific notes that inflate token cost without adding clarity.
Suggestions
Split the Discovery Timeout/Daemon Configuration and Extension Setup and Repair sections into reference files (e.g. references/daemon-config.md, references/setup.md) and keep SKILL.md as a route + readiness overview with one-level-deep pointers.
State the 'MCPORTER_CHROME_DEVTOOLS_RELAY_POLICY=require' prefix once as a standing convention for all example commands instead of repeating it on every line.
Collect version-sensitive notes (mcporter 0.13.10 fallback behavior, older/newer client timeout defaults) into a single 'Version-dependent behavior' or 'Legacy/older clients' section so the main flow stays timeless.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and almost entirely non-obvious operational detail, but it could be tightened: the long 'MCPORTER_CHROME_DEVTOOLS_RELAY_POLICY=require' prefix is repeated across roughly a dozen command lines, the relay-policy rationale is explained more than once, and time-sensitive material ('In mcporter 0.13.10', 'older clients', 'newer Chrome-specific outer defaults') sits inline rather than in a versions/deprecated section. This matches 'mostly efficient but includes some unnecessary explanation or could be tightened' (3), short of the 'minor instances' anchor (4). | 3 / 5 |
Actionability | Fully executable, copy-paste-ready mcporter commands with real arguments and output flags, a concrete JSON config block to merge into the canonical definition, and specific failure signatures ('network-error' at the wrong endpoint, 'browser_owner_conflict', stale uid reports, blocking 'Allow remote debugging?' prompts) each paired with a remedy. This matches 'fully executable; copy-paste ready code or commands; specific examples cover the common cases'. | 5 / 5 |
Workflow Clarity | The route is an explicit ordered fallback chain, setup/repair is a concrete checklist, and the 'Fail-Closed Readiness Proof' gives a five-condition numbered validation gate with commands. Error-recovery feedback loops are strong and explicit: a relay-policy error means report/repair instead of retrying, a stale uid means re-snapshot rather than retry, and an empty page list triggers a defined diagnosis sequence instead of escalation. This matches the top anchor including checklists and feedback loops. | 5 / 5 |
Progressive Disclosure | The single SKILL.md is well-sectioned with clear headers, but there are no bundle files at all (references/, scripts/, assets/ are absent) and the ~290-line body inlines material that clearly belongs in separate files — daemon and timeout configuration, extension setup/repair procedures, and the legacy fallback. This matches 'some structure but could be better organized; content that should be separate is inline' (3), above the unstructured anchor (2) but below the 'appropriately split' anchors (4-5). | 3 / 5 |
Total | 16 / 20 Passed |