CLI conventions for agents and CI

• Prefer --json for agent and CI calls. Server-mode render, preview, and play do not provide ordinary JSON output; preview --selection --json and preview --context --json are query-mode exceptions.
• doctor --json always exits zero. Gate on its payload:

bash npx hyperframes doctor --json | jq -e '.ok' >/dev/null

• Non-TTY mode is automatic and scaffolds the centered blank. Pass --example only to start from a named example. Use --non-interactive to force flag-only mode on a TTY.
• Use one HYPERFRAMES_RUN_ID for all commands in the same verification loop.
• When disk is tight, run npx hyperframes clean (--dry-run to list first); it removes what dead renders left and idle caches that rebuild themselves, never outputs, sources or anything a running render uses. Write disposable QC frames to a temp dir, then remove only that owned directory. Managed snapshot runs retain the newest 10 plus current/latest/rendered evidence; active captures and runs younger than 24 hours are protected. Copy intentional before/after evidence outside managed storage before it ages out. Never delete .hyperframes wholesale: it also holds project history.
• Use --strict, --strict-all, and --strict-variables when the corresponding warnings, variables, or CI conditions must gate the render.
• For shader_unbounded_cosine_pow, clamp the cosine to [0, 1] before pow(1 - cosine, exponent); a lower-only max is insufficient. Then preview and export on the affected GPU to confirm. Lint only recognizes static shader expressions, and software rendering can miss hardware-specific invalid pixels.
• JSON paths redact the home directory as $HOME; do not try to reverse the redaction.
• When a hosted cloud project approaches or exceeds the 200 MB upload limit, use cloud render --dry-run --json and follow the .hyperframesignore investigation in references/cloud.md. Never ignore an asset merely because it is large.
• Never render merely because checks pass. Pause at the final preview and wait for approval.