Workjournal

Delete a journal and all its entries. Irreversible. Owner-only. Pass workspace and journal slugs.

Delete a prompt. Cascades to remove it from any journal pool and resets `selected_prompt_id` on any journals selecting it back to the system default. Requires `prompt_templates` (Pro).

Delete a tag from the workspace registry. Cascades to remove the tag from every entry in every journal the tag was assigned to. Not tier-gated.

Export a journal as JSON, Markdown, or CSV. Returns the raw export payload as text. Pass workspace and journal slugs plus the desired format.

Fetch the full body of a single workjournal entry by the `id` returned from `search`. The id format is `<workspace>/<journal>/<index>` (e.g. `acme/engineering-notes/42`). Returns a JSON object `{id, title, text, url, metadata}` where `text` is the entry's full markdown content (title + summary + body).

Get a single journal entry by its per-journal index. Pass workspace and journal slugs plus the entry index (e.g. 1, 2, 3 — shown as `#N` in `list_entries` output).

Fetch a single journal by its slug. Returns journal metadata including the active `prompt: { source, name, slug, body, used_by_count }` — `prompt.body` is always populated (system default when no custom prompt is assigned). Call this before `create_entry` and follow `prompt.body` as the writing instruction for the journal.

Fetch a single prompt by slug. Returns the full body.

Fetch a single tag from the workspace registry by name. With `journal_slug`, also reports whether the tag is currently assigned to that journal.

Fetch a single workspace by its slug. Slugs are short, hyphenated, globally unique strings (e.g. `acme`, `juliusz-personal`). Use `list_workspaces` first if you do not know the slug.

Return the N most recent journal entries with their full body content. Pass workspace and journal slugs.

List the file attachments in a journal — id, filename, type, size, content URL, and the indices of entries that reference each one. Read-only.

List contributors of a journal (users with access, including the owner). Pass workspace and journal slugs.

List recent journal entries for a journal. Returns slim entries (no body). Use `last_entries` or `get_entry` to read entry content. Pass workspace and journal slugs.

List invitations for a journal (includes pending, accepted, and revoked). Pass workspace and journal slugs.

List the journals in a workspace that the signed-in user has access to. Pass the workspace slug from `list_workspaces`. Each journal returned has a `slug` you then pass to `get_journal`, `list_entries`, etc. Pass `shared-with-me` as the workspace slug to list journals shared with the user from workspaces they don't own.

List the workspace's prompts. Returns name, slug, and updated_at for each. Workspace owner only — non-owners get an empty list.

List tags in a workspace. Without `journal_slug`: returns the full workspace registry. With `journal_slug`: filters to tags assigned to that journal (the set entries in that journal can carry). Read-only — available to any workspace member (Free or paid). Each tag has `name` and `description`; with `journal_slug`, only assigned rows are listed. Tier gating applies only to create / assign / write operations.

List the workspaces the signed-in user owns or belongs to. Returns each workspace with its `slug` — that slug is the first argument to every other tool (e.g. `list_journals(workspace_slug)`). Most users have a single personal workspace named after their account.

Rename a journal (display name only). The slug stays the same, so existing links keep working. Use `set_journal_slug` if the user wants to change the URL slug instead.

Search across the user's accessible workjournal entries (their own workspaces' journals plus journals shared with them). Returns a JSON object `{results: [{id, title, text, url}]}` where each `id` is a handle of the form `<workspace>/<journal>/<index>` you can pass to `fetch` for the full entry body. Stable within a session, but not across workspace or journal slug renames (`set_journal_slug` breaks old ids the same way it breaks old web URLs). Use as the primary discovery tool when the user asks about something they've previously logged. Uses Postgres websearch syntax: bare words match anywhere, double-quoted phrases match exactly.

Search journal entries in a journal. Returns slim entries — follow up with `get_entry` for the full body. Pass workspace and journal slugs.

Set the journal's selected prompt. The target prompt must be in the journal's pool (or the global default — pass `prompt_slug = 'default'` to select it). Requires `custom_prompt`. Affects only the active selection — the journal's pool membership is untouched.

Semantic (RAG) search over a journal — ranks markdown-aware chunks of every entry by cosine similarity against the query embedding. Use this when the user asks about a concept rather than an exact keyword: 'find anything we decided about retry semantics', 'what did we say about pgvector'. Returns the matching CHUNKS (with their parent entry pointer + similarity score), not whole entries — follow up with `get_entry(workspace_slug, journal_slug, entry_index)` to read the full body. Plus+ tier; Free workspaces get a tier-rejection error verbatim.

Change a journal's URL slug. Changing the slug breaks any existing links to this journal — old `<workspace>/<journal>` URLs will return 404. Surface this consequence to the user before calling. The display name is unaffected; use `rename_journal` for that.

Unassign a tag from a journal. Cascades to strip the tag from every existing entry in that journal. The tag itself stays in the workspace registry — other journals are unaffected. Not tier-gated.

Reset the journal's selected prompt to the system default. Equivalent to calling `select_journal_prompt` with `prompt_slug = 'default'`. Requires `custom_prompt`. The journal's pool is untouched.

Update a journal entry by its per-journal index. All content fields are optional — only fields you pass are changed; the others (and `created_at`, `index`) are unchanged. At least one of `title`, `summary`, `body`, or `tags` must be provided. Use this to append verification notes, fix a typo, annotate an entry post-hoc, or replace the tag set — preferred over delete+recreate, which loses the original timestamp. To use `tags`, first call `list_tags(workspace_slug, journal_slug)`; unknown names are rejected with the valid options in the error.

Update a journal's display name and/or description in one call. Both fields are optional — only fields you pass are changed. Pass an empty string for `new_description` to clear it. At least one of `new_name` / `new_description` must be provided. Use `set_journal_slug` to change the URL slug (separate tool because it 404s old links).

Update a prompt's name and/or body. The slug is stable — renaming does not change the URL. Requires `custom_prompt`.

Rename a tag or replace its description. Renaming cascades to every entry in every journal the tag is assigned to. Description may be cleared. Not tier-gated — downgraded workspaces can still tidy up.

Upload a file (image or PDF, max 5 MiB) as an attachment in a journal by passing its bytes as base64. Works on every surface including the hosted remote server. Returns the attachment id and a content URL you can embed in an entry as `` or link to. Per-workspace storage quota applies by tier. On the local stdio server, prefer `upload_attachment_from_path` to avoid base64-encoding a file you already have on disk.