pub const CREATE_AFTER_LONG_HELP: &str = "Section / append / patch flag values:\n `--section KEY=VALUE`, `--append KEY=VALUE`, and `--patch KEY=OLD=>NEW`\n store the right-hand side as bytes verbatim. The CLI does NOT\n interpret backslash escapes \u{2014} `--section purpose=\"line1\\nline2\"`\n writes the literal two-character sequence `\\n` into the section\n body, not a newline.\n\n For multi-line section content, use `--from <FILE>` where FILE is a\n JSON payload matching the MCP `memstead_create` / `memstead_update` shape.\n The JSON parser de-escapes `\\n`, `\\t`, etc. before the engine\n sees the value, so a JSON-quoted `\"line1\\nline2\"` round-trips as\n two lines on disk.\n\nSlug derivation:\n The entity slug derives from the title in five steps:\n 1. NFC-normalize (combining sequences fold to precomposed form);\n 2. Unicode case-fold to lowercase;\n 3. rewrite each whitespace character to \'-\';\n 4. drop every character that is not Unicode alphanumeric and not \'-\';\n 5. collapse hyphen runs, trim leading/trailing hyphens.\n\n The mutation entry refuses titles where step 4 would drop any\n character, or where the pipeline output is empty (whitespace- or\n hyphen-only input). Errors carry a `proposed_slug` recovery hint\n so a sanitised retry is mechanical.\n\n The title body is stored as-sent (byte-form preserved); slug bytes\n derive from the NFC-normalised form. An NFD-spelled title therefore\n produces an NFC-spelled slug \u{2014} the two byte forms are semantically\n equivalent and compare equal under NFC normalization.\n\n Pre-gate entities (created before this stricter rule landed) remain\n readable. The gate runs at mutation entry only \u{2014} it does not\n retroactively reject entities loaded from disk.";Expand description
Combined --help epilog for memstead create: the section-bytes-
verbatim note followed by the title→slug pipeline description.
clap’s after_long_help takes a single string, so the two
epilogs are pre-concatenated here.