# Scrivener Inspect > When the user wants to open and explore a Scrivener project read-only — see its structure, read a chapter or scene, or locate a document. Use when the user says "show the outline", "what's in this project", "read this chapter/scene", "what's the synopsis of", "find the document about", or "list the binder". For converting/exporting content out, see scrivener-extract; for full-text/field search, see scrivener-search; for word counts and reading-time, see scrivener-stats. - Skill: `donnfelker/scrivener-inspect` (Agent Skill, multi-file: 3 files) - Install (CLI): `npx skillmds@latest add donnfelker/scrivener-inspect` - Raw SKILL.md: https://api.skillmd.com/api/skills/donnfelker/scrivener-inspect/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Docs & Writing - Author: donnfelker (https://skillmd.com/u/donnfelker) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/donnfelker/scrivener-inspect --- # Scrivener Inspect You help the writer open and explore a Scrivener 3 (`.scriv`) project **read-only**: show the binder outline, read a document's text/synopsis, and find an item by title, UUID, or type. Nothing here mutates the project, so no backup, snapshot, or "close Scrivener first" warning is required — but reads are most reliable when the project isn't mid-cloud-sync. Use this skill to *understand* a project before any editing skill acts on it. Compound requests usually start here: inspect → then hand a UUID to a write or export skill. ## When to use - "Show me the outline / list the binder" → `outline` - "Read this chapter / what does this scene say / what's the synopsis of X" → `read ` - "What's in this project / where's the document about Y / find the scene where…" → `find ` ## Commands All three are read-only. Output is JSON by default; add `--format text` for a human-readable view. Always pass `--project` pointing at the `.scriv` package (or its `.scrivx`). ### outline — the binder tree ```bash python3 ${CLAUDE_PLUGIN_ROOT}/tools/scrivener/cli.py outline --project "" python3 ${CLAUDE_PLUGIN_ROOT}/tools/scrivener/cli.py outline --project "" --format text ``` Returns the nested binder: each node has `uuid`, `title` (adaptive — see below), `type`, `label`, `status`, `include_in_compile`, and `children`. This is the map you read first to get the UUIDs other commands need. ### read \ — one document's content ```bash python3 ${CLAUDE_PLUGIN_ROOT}/tools/scrivener/cli.py read --project "" ``` Returns `uuid`, `title`, `type`, `include_in_compile`, `label`, `status`, `synopsis`, `has_synopsis_image`, the document `text` (RTF converted to plain text), and a `constructs` map flagging rich content. Pass a **UUID** when a title is ambiguous (titles are not unique). Folders and empty documents have no body text, so `text` and `constructs` will be `null` — that is normal, not corruption. ### find \ — locate items ```bash python3 ${CLAUDE_PLUGIN_ROOT}/tools/scrivener/cli.py find "" --project "" ``` Matches a **title substring** (case-insensitive), an **exact UUID**, or an **exact Type** (e.g. `Folder`, `Text`, `DraftFolder`). Returns `uuid`, `title`, and `type` for every match. Use it to turn a fuzzy "the chapter about the storm" into a concrete UUID, then `read` it. `find` is title/UUID/type only — for full-text or field search (text/synopsis/notes), use scrivener-search. ## Adaptive titles A binder item without a `` still gets a usable display title. The toolkit derives one in order: the document's synopsis first line → the body text's first line → `"Untitled"` (each truncated to ~60 chars). So an untitled scene shows up in `outline`/`read`/`find` with a meaningful, derived title rather than a blank. Treat these as *display* titles, not stable identifiers — address such items by **UUID**. ## What `read` returns A single `read` gives you the whole picture of one item: - **text** — the body, RTF→plain text (folders/empty docs return `null`). - **synopsis** + **has_synopsis_image** — the corkboard card text, or a flag that the synopsis is an image instead. - **label / status / include_in_compile** — the item's metadata (label/status are IDs; resolve names via scrivener-metadata's `labels`/`statuses`). - **constructs** — boolean flags for rich content embedded in the RTF: `comments_or_footnotes`, `style_markers`, `images`, `tables`, `hyperlinks`, `lists`. These tell you *before any edit* whether a document is "rich." Any `true` flag means a regenerative rewrite would lose data, so a downstream edit must be token-preserving or explicitly destructive (see scrivener-edit). ## Finding the Draft The Draft (manuscript) is identified by **Type `DraftFolder`**, never by title — a user can rename it ("Manuscript", "Novel", etc.). To locate it without guessing the name: ```bash python3 ${CLAUDE_PLUGIN_ROOT}/tools/scrivener/cli.py find "DraftFolder" --project "<path.scriv>" ``` Similarly, research lives under `ResearchFolder` and deleted items under `TrashFolder`. Reading the whole manuscript in order means walking the `DraftFolder`'s `children` from `outline`. ## Reference - `references/reading-items.md` — how `outline`/`read`/`find` output is shaped; UUID vs title addressing; reading notes and synopsis; and the meaning of every `constructs` flag. ## Toolkit support | Command | Status | |---------|--------| | `outline` | ✅ available now | | `read <id>` | ✅ available now | | `find <query>` (title substring / UUID / type) | ✅ available now | | Read notes, dates, bookmarks, links in `read` output | 🔜 planned (use scrivener-extract / scrivener-format today) | ## Related skills - **scrivener-extract** — convert/export content out (RTF→Markdown/text, bulk export of one item, a subtree, or the whole Draft). - **scrivener-search** — full-text and field search (text/synopsis/notes), beyond `find`'s title/UUID/type matching. - **scrivener-stats** — word counts, characters, reading-time and page estimates, writing history, and word frequency. - **scrivener-format** — the `.scriv` package format reference (binder schema, data files, the RTF constructs `read` flags).