common-operations

name

common-operations

description

This skill should be used when the user wants to do knowledge-base work in Outline — "create a document", "search Outline", "update a doc", "move a document to a collection", "create a collection", "share a document", "invite users to Outline", "star a document", "comment on a doc", or any everyday Outline operation. Provides plain-language workflows that drive the REST API and route to the exact methods.

Outline Common Operations

Plain-language playbooks for everyday Outline work. Each one drives the REST API. For exact method signatures and every field, open the api-reference skill's matching references/*.md file (named in each workflow).

Before anything: ensure access

If OUTLINE_API_KEY / OUTLINE_API_URL aren't confirmed this session, run the setup skill first (one auth.info call). Every call below assumes both are set and uses these headers:

-H "Authorization: Bearer ${OUTLINE_API_KEY}" -H "Content-Type: application/json" -H "Accept: application/json"

Define them once for brevity in the recipes below:

OUT="${OUTLINE_API_URL%/}"
H_AUTH="Authorization: Bearer ${OUTLINE_API_KEY}"

The golden rules (why workflows look the way they do)

1. Everything is POST ${OUT}/<method> with a JSON body. No GETs, no path params. An action with no inputs still needs -d '{}'.
2. Read .data. The useful payload is always under .data in the response envelope.
3. Resolve titles to ids once. Users say "the Welcome doc"; the API wants a UUID or urlId. Resolve with documents.search / documents.search_titles / collections.list and keep the id.
4. Confirm before destructive actions. documents.delete (trash), permanent: true, documents.empty_trash, collections.delete, users.delete/suspend, and shares.revoke are high-impact — show the user what will change first.

Workflow: find a document by title

DOC_ID=$(curl -s -X POST "${OUT}/documents.search_titles" -H "$H_AUTH" -H "Content-Type: application/json" \
  -d '{"query":"Welcome"}' | jq -r '.data[0].id')

Use documents.search instead when you need full-text matches with snippets/ranking. (→ documents.md)

Workflow: create a document

# Publish under a collection (collectionId OR parentDocumentId is required to publish)
curl -s -X POST "${OUT}/documents.create" -H "$H_AUTH" -H "Content-Type: application/json" \
  -d "{\"title\":\"Welcome\",\"text\":\"# Hello\\n\\nWelcome aboard.\",\"collectionId\":\"${COLLECTION_ID}\",\"publish\":true}" \
  | jq '.data | {id, title, url}'

• Draft: omit publish (or set false); publish later with documents.update {"id":"…","publish":true}.
• Child document: pass parentDocumentId instead of (or with) collectionId.
• From a file (markdown/docx/csv/html): use documents.import (multipart). (→ documents.md)

Workflow: update a document

# Full replace of the body
curl -s -X POST "${OUT}/documents.update" -H "$H_AUTH" -H "Content-Type: application/json" \
  -d "{\"id\":\"${DOC_ID}\",\"text\":\"# New body\"}" | jq '.data | {id, title}'

• Append / prepend without resending the whole doc: add "editMode":"append" (or prepend) with text.
• Surgical edit: "editMode":"patch" with findText (the existing text to replace) and text (the replacement). (→ documents.md)

Workflow: organize — move, archive, trash, restore

1. Move to another collection/parent: documents.move {"id","collectionId"?, "parentDocumentId"?, "index"?}.
2. Archive (hide, keep searchable): documents.archive {"id"}; reverse with documents.restore.
3. Trash (recoverable 30 days): documents.delete {"id"}; restore with documents.restore. Add "permanent":true to destroy immediately (confirm first).
4. Empty trash (admin, irreversible): documents.empty_trash — confirm. (→ documents.md)

Workflow: create a collection and add documents

1. Create — collections.create {"name","description"?,"permission"?,"color"?,"icon"?} (permission is read or read_write for the default workspace access). (→ collections.md)
2. Add docs — create documents with that collectionId, or documents.move existing ones in.
3. See the tree — collections.documents {"id"} returns the nested navigation structure.

Workflow: share a document publicly

SHARE_ID=$(curl -s -X POST "${OUT}/shares.create" -H "$H_AUTH" -H "Content-Type: application/json" \
  -d "{\"documentId\":\"${DOC_ID}\"}" | jq -r '.data.id')
# Shares start unpublished — publish to make it accessible without login
curl -s -X POST "${OUT}/shares.update" -H "$H_AUTH" -H "Content-Type: application/json" \
  -d "{\"id\":\"${SHARE_ID}\",\"published\":true}" | jq '.data | {id, url, published}'

Revoke with shares.revoke {"id"}. (→ sharing-access.md)

Workflow: manage people & permissions

1. Invite — users.invite {"invites":[{"email":"[e-mail]","name":"Alice","role":"member"}]}. (→ users-groups.md)
2. Find / list — users.list {"query":"alice"} or {"filter":"active"}.
3. Change role — users.update_role {"id","role":"admin|member|viewer"} (admin only).
4. Suspend (reversible, preferred over delete) — users.suspend {"id"}; reverse with users.activate. Confirm first.
5. Grant collection access — collections.add_user {"id":"<collectionId>","userId","permission":"read_write"}, or by group with collections.add_group. Per-document access uses documents.add_user / documents.add_group.

Workflow: engage — star, comment, view counts

• Star a doc/collection for the sidebar: stars.create {"documentId"} or {"collectionId"}.
• Comment: comments.create {"documentId","text":"…"}; reply with parentCommentId; anchor inline with anchorText. (→ comments-stars-views.md)
• View counts for a doc: views.list {"documentId"}.

Workflow: report & audit

1. Most-recent / by collection — documents.list {"collectionId"?, "sort":"updatedAt","direction":"DESC","limit":25}.
2. Activity insights for a doc (Business/Enterprise, must be enabled) — documents.insights {"id","startDate"?,"endDate"?}.
3. Audit trail — events.list {"name":"documents.create","auditLog":true} filtered by actorId/documentId/collectionId. (→ revisions-templates-events.md)
4. Paginate — walk {"limit":100,"offset":0}, then offset:100, … until a short page returns, or follow pagination.nextPath.

When a call fails (400/401/403/404/429), switch to the troubleshoot skill.