Conciseness

Nearly everything is tool-specific facts Claude cannot know (feature map, file locations, config merge rules) with no padding about general concepts — efficient overall. Not 5 because of duplicated routing: @reference/guide.md and @reference/workflows.md are each pointed to in 'Reference routing', again in 'How-To Guide', and guide.md a third time in the feature map, and on-disk locations appear in both 'Configuration Basics' and 'Where Things Live On Disk'. Not 3 because there is no unnecessary explanation, only redundancy that could be tightened.

Actionability

Concrete throughout: exact paths (`.mimocode/mimocode.json`, `~/.config/mimocode/mimocode.jsonc`), exact commands (`/vivid`, `/goal`, `mimo llm-server issue`), a copy-paste JSONC snippet with `$schema`, and a 7-step config-change procedure. Not 5 because of minor gaps — the validation command is never named ("Validate the parsed configuration with the narrowest relevant command") and the recent-model merge steps are deferred to a reference — which fit the 'concrete commands with minor gaps' anchor; not 3 because the guidance is specific and executable, not high-level hints.

Workflow Clarity

'Helping the User Configure' is a clearly sequenced 7-step procedure with explicit checkpoints: read the routed reference first, determine scope, inspect only exact candidates, redact secrets, edit minimally, validate, and report which files changed. Not 5 because there is no error-recovery branch (what to do when validation fails) and the validation command itself is unspecified; not 3 because validation and safety checkpoints are explicit and well placed, and the operations are not destructive/batch so no cap applies.

Progressive Disclosure

The body is a genuine overview: a feature map, config basics, and a dedicated 'Reference routing' section that maps each topic to exactly one @reference/*.md file, one level deep, with no nested references (the deepest hop is a single pointer like "(see @reference/guide.md)"). No bundle files were present in the evaluation directory to verify the paths exist, but the structure itself matches the top anchor: clear overview, well-signaled one-level references, easy navigation. Not 4 because every reference is unambiguously signaled by topic and none is buried.

Specificity

Concrete actions are named — "inspect existing config safely, make minimal changes, and verify them without guessing schema fields or model capabilities" — alongside a enumerated feature surface ("TUI or CLI commands, keybindings, terminal compatibility, rendering glitches, TUI lag, SSH or remote rendering, agent modes (build / plan)"). Not 5 because the 'what' side is thinner than the extensive trigger list (the core 'explain features' action is only implied by 'asks about MiMoCode itself'); not 3 because it goes well beyond 1-2 generic actions.

Completeness

Explicitly answers both: 'when' via "Use whenever the user asks about MiMoCode itself" plus two more trigger sentences, and 'what' via "inspect existing config safely, make minimal changes, and verify them". Not 4 because both what and when are explicit with concrete trigger phrases, mirroring the 5-anchor example structure.

Trigger Term Quality

Comprehensive natural-term coverage with synonyms and spelling variants users would actually type: "base URL/baseURL, API key/apiKey", "mimocode.json/jsonc", "TUI lag", "rendering glitches", "SSH or remote rendering", "how to enter or leave plan mode", "mimo llm-server". Not 4 because it explicitly includes variants and file extensions, matching the top anchor rather than leaving 'a few natural terms missing'.

Distinctiveness Conflict Risk

A clear niche — MiMoCode itself — with distinct triggers unlikely to fire for other skills: "mimocode.json/jsonc", "mimo llm-server", "MiMoCode", "agent modes (build / plan)". Not 4 because the unique product-specific tokens give minimal overlap risk with any generic skill.