Resolve effective StyleSeed context

Use the bundled scripts/resolve-context.mjs; do not hand-compose the rule stack.

1. Read STYLESEED.md when it exists. Confirm missing required selections with the user.
2. Keep the working directory at the user's project root. Invoke the script by its installed

path; do not cd into the skill directory. Prefer --from-lock STYLESEED.md; explicit flags override lock values.

1. Read the emitted .styleseed/effective-rules.md before building.
2. Preserve .styleseed/manifest.json. It records selections, source hashes, and the bundle hash

for updates and reproducibility.

1. Use --check to detect context drift without rewriting files.

bash
node <installed-ss-resolve>/scripts/resolve-context.mjs \
  --from-lock STYLESEED.md \
  --agent codex

Without a lock:

bash
node <installed-ss-resolve>/scripts/resolve-context.mjs \
  --agent claude \
  --grammar operations-console \
  --adapter product-ui \
  --domain saas \
  --page dashboard \
  --recipe enterprise-workbench \
  --palette cobalt-instrument \
  --profile swiss

Use --list to print supported IDs. --recipe auto maps the selected grammar to a maintained default; --palette auto maps that recipe to a contrast-verified semantic palette. Pass explicit values when the product needs a different morphology or color posture. The default output directory is .styleseed/ in the project root. For a project-local reference grammar, pass reference:<slug> and ensure .styleseed/rulesets/<slug>/RULESET.md exists.

Do not load llms-full.txt after a bundle resolves successfully. Load a larger source document only when the bundle names an unresolved ambiguity that requires it.