Skill Crossroads

Skill Crossroads delight
rubric v1.2 · deterministic

overall

Correctness & Structure 100 pass

Triggering & Discoverability 100 pass

Clarity & Instructions 100 pass

Token & Context Cost 91 1 warn

Safety & Security 89 1 warn

Verifiability & Maintainability 80 1 warn

Top fixes — ranked by score impact

⚠ SAFETY-03 No destructive auto-invocation

SKILL.md:22auto-invocable (no disable-model-invocation) → performs a destructive action ("publish")

Skill appears to "publish" but the model can auto-invoke it — a destructive action could fire without the user asking.

publish

Fix If this action is destructive/irreversible, set `disable-model-invocation: true` so it only runs when explicitly invoked.

⚠ TOKEN-02 Progressive disclosure

SKILL.mdeverything inlined in SKILL.md → 297 body lines, 0 supporting files

Large SKILL.md (297 lines) with no supporting files — all of it loads every time the skill fires.

Fix Move heavy reference material into supporting files and link to them, so the model loads it only when needed.

⚠ VERIFY-01 Evals or tests present

SKILL.md:293the skill can be verified → a prose verification step, but no runnable evals/ or tests/

Manual verification only — add eval prompts with expected outcomes to make it checkable.

## Verify

Fix Add an evals/ folder with 2–3 test prompts and the expected assertions.

★ scores higher than ≈61% of 208 public skills (deterministic rubric v1.2 sample, 2026-09)

Embed this badge

Always-fresh — it re-scans and updates on its own.

[![Skill Crossroads](https://skillcrossroads.com/api/badge/excatt/superclaude-plusplus/skills/delight.svg)](https://skillcrossroads.com/s/excatt/superclaude-plusplus/skills/delight)

Grade your own Claude Code skill — evidence-cited, file-and-line, free. Scan your own skill →