Slash Command4.4k repo starsupdated 4d ago

obsidian-export

The obsidian-export command generates a structured snapshot of an Obsidian vault in JSON or Markdown format, extracting metadata like titles, types, summaries, links, tags, and frontmatter from all notes. Use this command to create a machine-readable export that external tools, agents, or other AI systems can consume without needing to understand the vault's folder structure, and to maintain an operation log of export activities.

View source Repository: obsidian-second-brain

Install in Claude Code

Copy

mkdir -p ~/.claude/commands && curl -fsSL https://raw.githubusercontent.com/eugeniughelbur/obsidian-second-brain/HEAD/commands/obsidian-export.md -o ~/.claude/commands/obsidian-export.md

Then start a new Claude Code session; the slash command loads automatically.

Definition

obsidian-export.md

Use the obsidian-second-brain skill. Execute `/obsidian-export $ARGUMENTS`:

The optional argument is the format: `json` (default), `markdown`, or `okf`.

1. Read `_CLAUDE.md` first if it exists in the vault root
2. Read `index.md` for the full vault catalog

3. Build a structured export by scanning the vault:

   **For each note in every note folder present**, enumerating the vault's top-level note folders per `references/folder-map.md` (wiki-style: `wiki/`; Obsidian-style: `People/`, `Projects/`, `Ideas/`, `Knowledge/`, `Daily/`, ... whatever exists on disk), extract:
   - `path`: file path relative to vault root
   - `title`: note title (first heading or filename)
   - `type`: from frontmatter tags (entity, concept, project, daily, etc.)
   - `date`: from frontmatter
   - `status`: from frontmatter (if exists)
   - `summary`: first paragraph or first 200 characters of body
   - `links_to`: list of all outgoing `[[wikilinks]]`
   - `linked_from`: list of all incoming links (backlinks)
   - `tags`: all frontmatter tags
   - `frontmatter`: full frontmatter as key-value pairs

4. Output format:

   **JSON** (default):
   ```json
   {
     "vault": "Example Vault",
     "exported": "2026-04-07",
     "total_notes": 238,
     "notes": [
       {
         "path": "wiki/entities/Ada Lovelace.md",
         "title": "Ada Lovelace",
         "type": "entity",
         "summary": "CEO of Currentscale Labs...",
         "links_to": ["Currentscale Labs", "Tide Gateway"],
         "tags": ["entity", "person"]
       }
     ]
   }
   ```
   Save to `_export/vault-snapshot.json`

   **Markdown**:
   A flat markdown file with every note listed with its metadata and summary.
   Save to `_export/vault-snapshot.md`

   **OKF** (Open Knowledge Format - Google Cloud's vendor-neutral "folders of markdown" standard): do NOT build this by hand. Run the deterministic exporter from the skill root (its absolute path was given at session start as **Skill root**; substitute it for `SKILL_ROOT`):
   ```bash
   uv run --directory "SKILL_ROOT" scripts/export_okf.py --path "<vault path from _CLAUDE.md>"
   ```
   It writes an OKF v0.1 bundle to `_export/okf/`: every note becomes an OKF concept doc (frontmatter `type` [required] / `title` / `description` / `resource` [only when the note has a real source URL] / `tags` / ISO-8601 `timestamp`; `[[wikilinks]]` converted to relative-path markdown links), plus a generated `index.md` (progressive disclosure) and a copied `log.md`. The vault's richer AI-first body (incl. the `## For future agent` preamble) is preserved - OKF is minimally opinionated, so the extra content rides along. This makes the vault "OKF v0.1 compatible" without changing how it works natively.

5. Append to the operation log: if `Logs/` exists write `**HH:MM** - export | Vault snapshot exported (format, N notes)` to `Logs/YYYY-MM-DD.md`; otherwise append `## [YYYY-MM-DD] export | Vault snapshot exported (format, N notes)` to `log.md`

This file is the bridge between your vault and any other AI tool, automation, or agent. They don't need to know your folder structure. They read the snapshot.

---

**AI-first rule:** Every note created or updated by this command MUST follow `references/ai-first-rules.md` - `## For future agent` preamble, rich frontmatter (`type`, `date`, `tags`, `ai-first: true`, plus type-specific fields), recency markers per external claim, mandatory `[[wikilinks]]` for every person/project/concept referenced, sources preserved verbatim with URLs inline, and confidence levels where applicable. If that path does not resolve from your working directory, search upward for it; if you still cannot read it, say so before writing rather than producing a note that silently skips the rule. The vault is for future agent retrieval - not human reading.

**Anti-fabrication:** Search exhaustively before claiming any note, person, or file is absent - false absence is the most common failure mode - and never invent facts, entities, or dates (mark unknowns as `TBD`). See the anti-fabrication and search-completeness hard rules in `references/ai-first-rules.md`.