skill-authoringlisted

# skill-authoring House rules for building agent tooling in this repo: Claude Code-opinionated, open-standard compliant. A skill valid under the open Agent Skills standard is automatically valid in Claude Code (Claude Code makes every field optional), so the profile is: **write to the standard's rules as if mandatory, add Claude Code extensions only as progressive enhancement.** Verified against official docs 2026-07-01 (code.claude.com/docs/en/skills, agentskills.io/specification, platform.claude.com agent-skills best-practices). Re-verify before trusting version-sensitive details. ## The profile ### Core (open standard, treat as mandatory) | Field | Rule | |---|---| | `name` | lowercase a-z 0-9 hyphens, 1-64 chars, **must match the directory name**, no leading/trailing/consecutive hyphens, no "anthropic"/"claude" | | `description` | 1-1024 chars hard limit, **600-800 is the house target**, **third person**, states what the skill does AND the concrete conditions that should trigger it, no XML tags. A plain scalar must contain no `: ` (colon followed by space) or the YAML fails to parse; use a spaced hyphen, or a `>-` folded block. | | `license` | optional, include when a skill may be shared | | `metadata` | optional string map; put custom keys (e.g. `version`) HERE, not top-level | | `allowed-tools` | optional, experimental in the standard | ### Claude Code extensions (use deliberately; no-ops elsewhere) `when_to_use`, `argument-hint`, `arguments`, `disable-model-invo