Adapts the upstream `skill-creator` skill (Apache-2.0,
JuliusBrussee/awesome-claude-skills @ commit 5380239) into a new
framework skill at `.claude/skills/write-skill/`. The upstream
flow shape (anatomy of a skill, progressive disclosure, 6-step
creation process) is preserved; the framework-specific shape and
the security patterns from the 2026-05 audit are baked in as
defaults so future skills authored through this flow inherit
the lessons rather than rediscovering them.

Substantial adaptations versus upstream:

- Renamed `skill-creator` → `write-skill` to match the
  framework's verb-prefixed naming convention.
- Frontmatter rewritten to the framework schema:
  `license: Apache-2.0` (exact string), `when_to_use` alongside
  `description`, SPDX header + placeholder-convention comment.
- Step 3 (initialisation) uses the adapted `init_skill.py` that
  scaffolds the framework's expected preamble: Adopter
  overrides, Snapshot drift, placeholder convention, SPDX
  header, plus conditional placeholders for the
  injection-guard callout and the Privacy-LLM gate-check.
- Step 5 (packaging) dropped — the framework distributes skills
  via the snapshot model, not zip artefacts. The upstream's
  `package_skill.py` and `quick_validate.py` are not included;
  validation is via the framework's existing
  `tools/skill-validator/`.
- New Step 5 (security checklist) — a hard walk-through of the
  nine prompt-injection-defence patterns from the gist audit.
  The patterns live in
  `.claude/skills/write-skill/security-checklist.md`. This is
  the load-bearing adaptation: it ensures any new skill
  written through this flow inherits the audit's lessons.

Attribution per ASF licensing-howto:

- LICENSE.txt copied verbatim from upstream into the skill
  directory.
- Project root NOTICE gets a "Third-party content" entry
  crediting Julius Brussee and the upstream repo.
- SKILL.md § "Provenance" pins the exact upstream commit and
  enumerates the adaptations.

Generated-by: Claude Code (Claude Opus 4.7)