Conciseness

Efficient and competence-assuming — it never explains what a JWT, an OpenAI-compatible endpoint, or a grep is; every section (scoping test, marker, naming map, rules) carries load. A few rhetorical flourishes could be trimmed ("Treat every touch as work on a product that will one day live on its own", "The grep is the index", the closing blockquote), keeping it just below the lean-every-token-earns-its-place anchor.

Actionability

Concrete and executable: a copy-paste grep ("grep -rn 'GRIDA-GG' editor packages desktop"), exact marker comment forms for each surface, a complete identifier table, and a dual-marker example. Minor gaps: the dependency-direction rule ("The gateway must not import desktop/renderer code") states the invariant but gives no verification command, and the working rules reference paths without a check step.

Workflow Clarity

"Working on GG" is a clear 4-step sequence with rationale, and the security crossing has an explicit ordering checkpoint ("runs the `security` review **first** ... then this skill") plus a stated invariant. This is a code-organization skill, not a destructive/batch operation, so the missing validation/verification step (e.g. re-running the grep to confirm tagging) is a minor gap rather than a cap.

Progressive Disclosure

With no bundle files present, all content is appropriately inline in well-labeled sections; every outward reference is one level deep and clearly annotated with its role — canonical spec URL, SECURITY.md boundary, and each companion skill ("the GRIDA-SEC-006 half", "the credit ledger GG spends", "the two hosts"). Navigation is easy with a clean overview-to-detail flow.

Specificity

Enumerates several concrete components — "the scoped-token mint, the OpenAI-compatible + native gateway endpoints, the `gg` client provider, and the desktop wiring that spends org credit without a user key" — giving comprehensive coverage of the surface, but describes the pattern's components rather than the actions performed (tagging, naming, placement), a minor gap versus the top anchor.

Completeness

Clearly answers both questions: the "what" is the enumerated four-surface working pattern for the metered no-BYOK gateway, and the "when" is explicit with concrete triggers — "Use when adding or touching any GG file, the `gg` provider kind, the `gg:ai` token audience, or deciding whether code belongs to the gateway surface." Voice is third-person throughout.

Trigger Term Quality

Good keyword coverage of the terms a repo user would naturally say: "Grida Gateway (GG)", "GG file", "the `gg` provider kind", "the `gg:ai` token audience", "gateway surface", "no-BYOK", "org credit". A few natural variants (e.g. "hosted AI", "grida.gg") are absent, so it falls just short of the comprehensive-synonyms anchor.

Distinctiveness Conflict Risk

A clear niche with distinct markers (`GRIDA-GG`, `gg:ai`, `gg` provider) and explicit delegation of the security half to the `security` skill, but the shared `GRIDA-SEC-006` boundary and the broad "deciding whether code belongs to the gateway surface" trigger leave minor overlap risk with the `security` and `ee-billing` companions — mostly distinct rather than minimal-conflict.