Krosoft Wiki
Auto Light Dark
Auto Light Dark

Tools

Tool contracts for direct integrations. Use the connection’s tools/list for current schemas and available operations. Examples are argument objects; replace sample IDs, versions, cursors and space keys with values from your connection.

Common contract

All calls run as the authenticated caller. Native permissions, MCP mode, OAuth scopes and administrator policy determine access. Missing, inaccessible or policy-excluded content is not disclosed. See Security and permissions.

Types and pagination

IDs are positive decimal strings of at most 19 digits within signed 64-bit range. Versions are integers from 1 through 2,147,483,646. Unknown fields, nulls, wrong types and out-of-range values are rejected. Text lengths use UTF-16 code units.

Offset lists use start (0 through 10,000; default 0) and limit (1 through 25; default 10). Related-content lists return results, start, limit, hasMore and, when available, nextStart. Follow that next offset, even after a short or empty filtered batch. paginationLimitReached: true marks the offset ceiling. Search has its own pagination described below.

Navigation and trash lists instead use opaque cursors, incomplete, optional nextCursor and paginationLimitReached. Follow returned cursors unchanged; a short or empty batch can be incomplete. Lists are not snapshots and expose no hidden totals. Limits and node-affinity requirements are in Compatibility and limitations.

Results and errors

Tool results are JSON encoded in MCP content[0].text. Check isError; HTTP 200 alone does not establish success. Results identify the caller with executedAs.

Errors and retries

Failure fields and retry rules

Failures include error, errorCategory and requestId. invalid_arguments covers validation and preflight rejections; permission_or_unavailable combines missing/inaccessible content and permission/policy denials; native optimistic conflicts use version_conflict. Configuration, timeouts, I/O and unexpected failures use sanitized categories without credentials, native exception details or local paths.

Mutations also report mutationOutcome and retryGuidance. Known validation, permission or conflict rejections are not_applied; unexpected failures may be unknown. Multi-step copies can be partial_or_unknown. Follow the returned outcome and guidance. A failed or lost response can follow a successful save: inspect native state before retrying. The app never retries mutations automatically; creates, replies and uploads have no client idempotency key. See MCP for Confluence Data Center - Troubleshooting for diagnostics.

Safe editing and storage format

Read the current complete storage body and version before editing. Preserve unrelated content, macros, links, tables and tasks. confluence_update_page replaces the complete title/body; confluence_update_section replaces only the selected section body. Supply the current version, not its successor. After a conflict, read again and reconcile; after saving, verify the persisted result.

Page replacement boundary

Write limits and storage rules

Full-page and section updates require both the existing page and resulting page to fit 50,000 characters, even if the replacement is shorter. Completing a chunked read does not lift this write limit. Use Confluence’s editor for larger pages. Create bodies and comment bodies also have a 50,000-character limit; comment edits refuse oversized existing comments.

Write bodies use complete Confluence storage XML, never Markdown or wiki markup. New comments and section edits validate complete XML fragments with ac and ri namespace bindings; other namespaces must be declared. DTDs and external entities are rejected. Native validation does not prove that an edit preserved unrelated content. The small update examples below represent entire replacement bodies, not patches.

The request envelope and byte limits can be tighter for escaped content. See Compatibility and limitations for request and deployment limits.

confluence_get_current_user

Identify the connected account and effective MCP policy.

Arguments, result and example

No arguments; pass {}.

Result: executedAs, userKey, spaceScopeMode, allowedSpaces, blockedSpaces and operationMode.

Space modes are native_permissions, allowlist, blocklist or allowlist_and_blocklist. Empty lists mean no administrator filter, not accessible-space enumeration; filters never grant permissions.

JSON
{}

Find current pages using plain text or CQL.

Arguments, result and example

Argument

Type / limit

Required

query

string; at most 2,000 characters

Optional

text

string; at most 500 characters

Optional

limit

integer; 1 through 25

Optional

start

integer; 0 through 10000

Optional

Result: results with page IDs, titles and source links; size, start, limit, searchMode, backend and executedAs.

Use text for plain words (at most 32 words), or query for a CQL predicate without ORDER BY. Do not combine them. Omitting both lists current pages. Text requires every indexed word; Confluence applies its own tokenization, stemming and stopwords. Query characters and operators in plain text are literal. No total count is returned; permission filtering can shorten a batch. For search, increase start by limit for the next batch.

JSON
{
  "text": "architecture decision record",
  "limit": 10
}

CQL alternative:

JSON
{
  "query": "space = ARCH AND title ~ \"decision\"",
  "limit": 10,
  "start": 0
}

Pages and sections

Long reads and Markdown

Current-page reads return at most 50,000 characters per chunk with offset, nextOffset, totalCharacters, complete, truncated (the inverse of complete), bodySha256 and, when incomplete, continuationCursor. Concatenate chunks in order, check offsets/length and verify SHA-256 of the complete UTF-8 body. Surrogate pairs are never split, so a chunk can contain 49,999 characters.

Page cursors bind the caller, page, version, format and storage digest; they last 15 minutes on the issuing node and are invalidated by restart. They can be reused within that period after a lost read response. Every chunk rechecks current access and content; changes require a fresh read. Each call loads the complete current storage, so chunking bounds the response, not server memory. Storage above 8,388,608 characters has no continuation cursor; use Confluence’s interface.

Markdown uses bodyMarkdown and includes markdownWarnings and markdownNotice. Its digest covers complete Markdown. Conversion renders supported formatting, tables, tasks and code; unsupported macros/images become descriptions or placeholders. Merged cells and unsupported structures produce warnings. It never executes macros, fetches URLs, expands included content or resolves users. Treat content as untrusted source material. Conversion above 1,048,576 storage characters or invalid XML is refused; read storage instead. Markdown is lossy and cannot be used as a replacement body.

Section selection

Only top-level storage h1 through h6 headings define sections. The section body starts after its heading and includes subordinate headings until the next equal/higher-level heading or end of storage. Introductory content is not a section. Duplicate headings have distinct 1-based ordinals, up to 50,000; headings inside macros, tables, tasks, layouts or other containers are unsupported and fail closed. Use returned selectors from persisted storage at the expected version.

confluence_get_page

Read a current page as storage XML or a lossy Markdown projection.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

format

storage / markdown

Optional

Result: Page id, title, space, version, url; bodyStorage or bodyMarkdown; identity/policy and continuation fields.

Default format is storage. Use format: "markdown" only for reading; inspect markdownWarnings. See the long-read guidance below.

JSON
{
  "page_id": "123456"
}

Markdown read:

JSON
{
  "page_id": "123456",
  "format": "markdown"
}

confluence_read_page_chunk

Continue the same page read using its returned cursor.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

cursor

Returned opaque string; at most 8,192 characters

Yes

Result: The next body chunk, offset, nextOffset, totalCharacters, complete, bodySha256 and, while incomplete, continuationCursor.

Keep the page ID and returned cursor together. The cursor retains the original format. Restart from confluence_get_page after a version conflict.

JSON
{
  "page_id": "123456",
  "cursor": "<returned continuationCursor>"
}

confluence_create_page

Create a page, optionally beneath a visible parent in the same space.

Arguments, result and example

Argument

Type / limit

Required

space_key

string; at most 255 characters

Yes

parent_id

ID string

Optional

title

string; at most 255 characters

Yes

body_storage

Storage XML; at most 50,000 characters

Yes

Result: id, version, url and executedAs.

Creation is not idempotent. Check for an existing saved page after a lost response before retrying.

JSON
{
  "space_key": "DEV",
  "title": "Integration notes",
  "body_storage": "<p>Created through MCP.</p>"
}

confluence_update_page

Replace a page’s complete title and storage body at its current version.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

title

string; at most 255 characters

Yes

body_storage

Storage XML; at most 50,000 characters

Yes

Result: id, resulting version, url and executedAs.

Supply the current version, without incrementing it. Space, ancestry and direct restrictions remain. Follow Safe editing and storage format below.

JSON
{
  "page_id": "123456",
  "version": 7,
  "title": "Integration notes",
  "body_storage": "<p>Complete revised body.</p>"
}

confluence_list_sections

Discover sections or read one section at an expected page version.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

section_index

integer; 1 through 50000

Optional

start

integer; 0 through 10000

Optional

limit

integer; 1 through 25

Optional

Result: Discovery: sectionIndex, level, headingStorage. Selected read: pageId, version, heading and bodyStorage, with truncated: false.

Select a returned section_index to read its body; do not combine a selector with start or limit. Section ordinals, not heading strings, identify sections.

JSON
{
  "page_id": "123456",
  "version": 7,
  "limit": 10
}

Selected read:

JSON
{
  "page_id": "123456",
  "version": 7,
  "section_index": 2
}

confluence_update_section

Replace only the selected section body, preserving its heading and surrounding storage.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

section_index

integer; 1 through 50000

Yes

body_storage

Storage XML; at most 50,000 characters

Yes

Result: id, resulting version, url and executedAs.

Use the exact ordinal discovered at the current version. Rediscover sections after a successful edit or conflict. Confluence may normalize storage when saving.

JSON
{
  "page_id": "123456",
  "version": 7,
  "section_index": 2,
  "body_storage": "<p>Revised decision.</p>"
}

Comments, labels and history

Related-content rules

These tools require current owning-page access, including for history. Comment reads can include permitted inline comments, but mutations support footer comments only. Each comment is checked for ownership/visibility; native per-comment edit/remove rules apply, including author-specific rules. Page edit permission alone is insufficient.

Only global labels are exposed. Labels contain 1 through 255 lowercase letters, digits, underscores or hyphens and start with a letter or digit. Mutations require native page edit and action-specific label permission; personal/system and unrelated labels remain unchanged. Historical reads never restore a prior revision.

confluence_list_comments

List comment metadata and reply relationships.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

start

integer; 0 through 10000

Optional

limit

integer; 1 through 25

Optional

Result: results with id, version, author, editor, optional modifiedAt/parentCommentId, and inline; no bodies.

JSON
{
  "page_id": "123456",
  "limit": 10
}

confluence_get_comment

Read a comment belonging to the specified page.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

comment_id

ID string

Yes

Result: comment metadata, bodyStorage and truncated.

JSON
{
  "page_id": "123456",
  "comment_id": "456"
}

confluence_add_comment

Add a footer comment or reply to a footer comment on the same page.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

parent_comment_id

ID string

Optional

body_storage

Storage XML; at most 50,000 characters

Yes

Result: id, version and executedAs.

Omit parent_comment_id for a new top-level footer comment.

JSON
{
  "page_id": "123456",
  "parent_comment_id": "456",
  "body_storage": "<p>Agreed.</p>"
}

confluence_update_comment

Replace a footer comment body at its current comment version.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

comment_id

ID string

Yes

version

Current integer version

Yes

body_storage

Storage XML; at most 50,000 characters

Yes

Result: id, resulting version and executedAs.

Use the comment version, not the page version. Oversized current comments are refused.

JSON
{
  "page_id": "123456",
  "comment_id": "456",
  "version": 2,
  "body_storage": "<p>Revised comment.</p>"
}

confluence_remove_comment

Remove a footer comment using native comment permissions.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

comment_id

ID string

Yes

Result: removed and executedAs.

Confluence determines how replies are handled. Inline comment mutations are unavailable.

JSON
{
  "page_id": "123456",
  "comment_id": "456"
}

confluence_list_labels

List global labels on a page.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

start

integer; 0 through 10000

Optional

limit

integer; 1 through 25

Optional

Result: results containing name and prefix.

JSON
{
  "page_id": "123456",
  "limit": 10
}

confluence_add_label

Add one global label without replacing unrelated labels.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

label

Global label; at most 255 characters

Yes

Result: label, action and executedAs.

JSON
{
  "page_id": "123456",
  "label": "architecture-decision"
}

confluence_remove_label

Remove only the named global label.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

label

Global label; at most 255 characters

Yes

Result: label, action and executedAs.

JSON
{
  "page_id": "123456",
  "label": "architecture-decision"
}

confluence_list_versions

List historical version metadata in descending order.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

start

integer; 0 through 10000

Optional

limit

integer; 1 through 25

Optional

Result: results, currentVersion; each record includes version, historical title, author, optional UTC modification time, message and metadataTruncated.

Deleted versions are skipped while advancing the offset. Messages are limited to 1,024 characters. Concurrent edits can shift offsets.

JSON
{
  "page_id": "123456",
  "limit": 10
}

confluence_get_version

Read one historical version under current page permissions.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

Result: metadata, bodyStorage and truncated.

Historical bodies stop at 50,000 characters and have no chunk continuation. This reads history; it does not restore a historical revision.

JSON
{
  "page_id": "123456",
  "version": 3
}

Attachments and transfers

File names, bytes and upload sessions

Inline read/upload/copy is limited to 32,768 decoded bytes (43,692 base64 characters), including zero-byte files. Base64 must be canonical and padded when required. Files above that limit need chunk tools. No arbitrary URL or server path is accepted or returned.

Filenames are nonblank simple names of at most 255 characters, without leading/trailing whitespace, separators, colon, double quote or controls; . and .. are invalid. MIME types are lowercase, without parameters, at most 127 characters. Declared MIME syntax is not content sniffing or malware scanning. Native site upload restrictions still apply.

Upload sessions bind the caller, page and declared metadata/version. Begin, append and commit on the same node using returned offsets. Access and replacement version are rechecked. Explicit new versions retain native history/editor attribution; inline upload and copy always require new names. Copy checks both endpoints and source attachment visibility/version, and does not copy history.

Sessions use private temporary storage and expire after 15 minutes. Commit, cancellation and shutdown remove pending files; crashes can require administrator cleanup. Large copies use the same disk quota. See Transfer limits for file, session and temporary-storage limits.

confluence_list_attachments

List attachment metadata on a page.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

start

integer; 0 through 10000

Optional

limit

integer; 1 through 25

Optional

Result: results with id, filename, mediaType, sizeBytes and version.

JSON
{
  "page_id": "123456",
  "limit": 10
}

confluence_get_attachment

Read an attachment of at most 32 KiB as base64.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

attachment_id

ID string

Yes

Result: attachment metadata, dataBase64 and truncated: false.

Larger files fail rather than returning partial binary; use confluence_read_attachment_chunk.

JSON
{
  "page_id": "123456",
  "attachment_id": "456"
}

confluence_upload_attachment

Create a new attachment filename from at most 32 KiB of base64 data.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

filename

string; at most 255 characters

Yes

media_type

string; at most 127 characters

Yes

data_base64

Canonical padded base64; at most 43,692 characters

Yes

Result: attachmentIds (one ID), filename, sizeBytes, version: 1 and executedAs.

An occupied filename is refused. For an explicit new version, use the upload session tools.

JSON
{
  "page_id": "123456",
  "filename": "note.txt",
  "media_type": "text/plain",
  "data_base64": "SGVsbG8K"
}

confluence_copy_attachment

Copy an attachment of at most 32 KiB to a new destination filename.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

attachment_id

ID string

Yes

destination_page_id

ID string

Yes

filename

string; at most 255 characters

Yes

Result: attachmentIds (one ID), filename, sizeBytes, version: 1 and executedAs.

Copies bytes and native media type, not attachment history. It does not move the source; an occupied destination name is refused.

JSON
{
  "page_id": "123456",
  "attachment_id": "456",
  "destination_page_id": "789",
  "filename": "copied-note.txt"
}

confluence_begin_attachment_upload

Start an upload for a new filename or an explicit attachment version.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

filename

string; at most 255 characters

Yes

media_type

string; at most 127 characters

Yes

size_bytes

integer; 0 through 104857600

Yes

sha256

64 lowercase hexadecimal characters

Yes

mode

create / new_version

Yes

attachment_id

ID string

Optional

version

Current integer version

Optional

Result: transferId, chunkBytes, maxTransferBytes, expiresAt and nextOffset: 0.

Use mode: "create" without attachment selectors, or mode: "new_version" with current attachment_id and version; its filename must match. Declare exact length and lowercase SHA-256.

JSON
{
  "page_id": "123456",
  "filename": "note.txt",
  "media_type": "text/plain",
  "size_bytes": 6,
  "sha256": "66a045b452102c59d840ec097d59d9467e13a3f34f6494e539ffd32c1bb35f18",
  "mode": "create"
}

confluence_append_attachment_chunk

Append bytes at the upload’s exact next offset.

Arguments, result and example

Argument

Type / limit

Required

transfer_id

UUID string returned by begin upload

Yes

offset

integer; 0 through 104857600

Yes

data_base64

Canonical padded base64; at most 43,692 characters

Yes

Result: The next upload offset (nextOffset).

Each chunk decodes to 1 through 32,768 bytes. Replayed or out-of-order chunks are refused. Use the returned offset; do not retry chunks blindly. Skip append for an empty file.

JSON
{
  "transfer_id": "11111111-1111-1111-1111-111111111111",
  "offset": 0,
  "data_base64": "SGVsbG8K"
}

confluence_commit_attachment_upload

Verify the complete length and digest, then publish the upload once.

Arguments, result and example

Argument

Type / limit

Required

transfer_id

UUID string returned by begin upload

Yes

Result: attachment metadata and mutationOutcome: "applied".

Incomplete uploads remain pending. Digest failure discards the upload; commit consumes it before native publication. A failed or lost response requires destination reconciliation.

JSON
{
  "transfer_id": "11111111-1111-1111-1111-111111111111"
}

confluence_cancel_attachment_upload

Discard the caller’s pending upload and temporary bytes.

Arguments, result and example

Argument

Type / limit

Required

transfer_id

UUID string returned by begin upload

Yes

Result: cancelled: true and executedAs; no native content is changed.

Expired, cancelled, consumed or foreign transfers are refused without revealing ownership.

JSON
{
  "transfer_id": "11111111-1111-1111-1111-111111111111"
}

confluence_read_attachment_chunk

Read a bounded byte range at an expected attachment version.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

attachment_id

ID string

Yes

version

Current integer version

Yes

offset

integer; 0 through 104857600

Yes

length

integer; 1 through 32768

Optional

Result: Attachment metadata, offset, nextOffset, complete and dataBase64.

Default length is 32,768 bytes. Decode each chunk before concatenating bytes; do not concatenate padded base64 strings. Verify binary integrity and restart if the native version changes.

JSON
{
  "page_id": "123456",
  "attachment_id": "456",
  "version": 2,
  "offset": 0,
  "length": 32768
}

confluence_copy_large_attachment

Copy a specified attachment version to a new destination filename.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

attachment_id

ID string

Yes

version

Current integer version

Yes

destination_page_id

ID string

Yes

filename

string; at most 255 characters

Yes

Result: attachment metadata and mutationOutcome.

The configured transfer limit applies. Rechecks source version and destination access; existing filenames are refused. No history is copied.

JSON
{
  "page_id": "123456",
  "attachment_id": "456",
  "version": 2,
  "destination_page_id": "789",
  "filename": "copied-file.pdf"
}
Navigation and move rules

Navigation checks every ancestry edge and refreshes page metadata; hidden paths fail closed and pages moved outside the requested root are omitted. Lists scan at most 100 native records per call and stop at a 10,000-record offset. Cursors bind caller, query, limit and node, expire after 15 minutes and fail after restart. These checks do not create a transactional snapshot.

Moves preserve page identity, storage, attachments and direct restrictions. Destination ancestry can change inherited access. Native move permissions, source/destination state, sibling state, cycles and the full bounded subtree must pass. Homepages are refused. Moves are limited to 25 pages/five levels and stay within one space; cross-space moves need Confluence’s interface. No permission administration is exposed.

Page-copy rules

Copy allows at most 25 pages and five levels. New titles combine the nonblank prefix (at most 100 characters) and original title, up to 255 characters; occupied titles fail preflight. Source ancestry must be visible and unrestricted; inaccessible descendants cannot be silently omitted. Cross-space copy requires both space policies and destination create permission; the destination cannot lie inside the source tree.

Choose attachments and global labels explicitly. Copy retains storage, original links/macros and their targets; it does not retarget links within the new tree. New pages inherit destination access. Comments, history and properties are excluded. Limits include 25 attachments and 25 labels per page and 10 MiB total attachment bytes per copy.

Each call copies at most one page. An ordinary in_progress result has isError: false and a new single-use copyCursor. Continue with only copy_cursor; add cancel: true to discard remaining work. Cursors bind caller/node and expire 15 minutes after the initial call; use node affinity. Cancellation retains saved pages. See Pending copy limits.

Source, destination and previously copied page state are rechecked. A failed attachment publication can leave the page and earlier attachments saved, stop the plan and return partial_or_unknown without another cursor. Outcomes include visible sourceId, destinationId when available, and applied, unknown, not_applied or not_attempted status. Records no longer visible are withheld without omitted counts, so outcomes are not an exhaustive audit. Inspect each returned destination and native attachments before manual recovery; never replay the whole tree automatically.

confluence_list_spaces

Browse accessible spaces and permitted homepages.

Arguments, result and example

Argument

Type / limit

Required

cursor

Returned opaque string; at most 8,192 characters

Optional

limit

integer; 1 through 25

Optional

Result: results with space key, name and type, and a homepage only when its ancestry is visible; traversal fields.

JSON
{
  "limit": 10
}

confluence_list_pages

Browse a page’s children or descendants.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

depth

children / descendants

Optional

cursor

Returned opaque string; at most 8,192 characters

Optional

limit

integer; 1 through 25

Optional

Result: Page id, title, space, version, parentId, position and source URL; traversal fields.

Default depth is children. Returned parent and position values supply move preconditions; position: -1 means unset.

JSON
{
  "page_id": "123456",
  "depth": "children",
  "limit": 10
}

confluence_get_ancestors

Read root-first permitted ancestry and the page’s own metadata.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

Result: Ancestors plus current page metadata, including parent and position.

Hidden or changing paths fail closed. Paths longer than 50 pages are refused.

JSON
{
  "page_id": "123456"
}

confluence_copy_page

Copy a page or bounded tree, one page per call.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Initial call

version

Current integer version

Initial call

destination_page_id

ID string

Initial call

destination_version

Current integer version

Initial call

title_prefix

string; at most 100 characters

Initial call

children

none / bounded

Initial call

max_pages

integer; 1 through 25

Initial call

max_depth

integer; 1 through 5

Initial call

attachments

exclude / include

Initial call

labels

exclude / include

Initial call

restrictions

reject

Initial call

links

preserve

Initial call

acknowledge_destination_access

boolean; must be true

Initial call

copy_cursor

UUID string returned by copy

Continuation

cancel

boolean

Optional on continuation

Result: complete, mutationOutcome, visible per-item outcomes, and, while in progress, copyCursor/expiresAt.

The initial call requires every initial field shown. Continue using only copy_cursor; optionally add cancel: true. Do not combine initial and continuation fields. See page-copy rules below.

JSON
{
  "page_id": "123456",
  "version": 7,
  "destination_page_id": "789",
  "destination_version": 3,
  "title_prefix": "Copy of ",
  "children": "none",
  "max_pages": 1,
  "max_depth": 1,
  "attachments": "exclude",
  "labels": "exclude",
  "restrictions": "reject",
  "links": "preserve",
  "acknowledge_destination_access": true
}

Continuation (use the latest returned cursor):

JSON
{
  "copy_cursor": "11111111-1111-1111-1111-111111111111"
}

To cancel remaining work, add cancel: true.

confluence_move_page

Move or reorder a page within its current space.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

current_parent_id

ID string; 0 for no parent

Yes

current_position

Integer position; -1 means unset

Yes

destination_page_id

ID string

Yes

destination_version

Current integer version

Yes

placement

child / before / after

Yes

sibling_id

ID string

Optional

sibling_version

Current integer version

Optional

sibling_position

Integer position; -1 means unset

Optional

acknowledge_destination_access

boolean; must be true

Yes

Result: page metadata, mutationOutcome and access notice.

The destination is the new parent. before/after placement also requires sibling_id, sibling_version and sibling_position for a child of that parent. child rejects sibling fields. current_parent_id: "0" denotes no parent. Re-read ancestry and positions after moving.

JSON
{
  "page_id": "123456",
  "version": 7,
  "current_parent_id": "321",
  "current_position": 0,
  "destination_page_id": "789",
  "destination_version": 3,
  "placement": "child",
  "acknowledge_destination_access": true
}

confluence_trash_page

Reversibly trash a current non-homepage without children.

Arguments, result and example

Argument

Type / limit

Required

page_id

ID string

Yes

version

Current integer version

Yes

Result: id, trashed: true and mutationOutcome.

Uses the expected current page version and native delete permission. No recursive trash or permanent purge.

JSON
{
  "page_id": "123456",
  "version": 7
}

confluence_list_trash

Browse trashed page metadata as a native space administrator.

Arguments, result and example

Argument

Type / limit

Required

space_key

string; at most 255 characters

Yes

cursor

Returned opaque string; at most 8,192 characters

Optional

limit

integer; 1 through 25

Optional

Result: Bounded metadata for restoration and traversal fields; no page bodies.

JSON
{
  "space_key": "DEV",
  "limit": 10
}

confluence_restore_page

Restore a trashed page beneath an explicit current parent in its original space.

Arguments, result and example

Argument

Type / limit

Required

space_key

string; at most 255 characters

Yes

page_id

ID string

Yes

version

Current integer version

Yes

destination_page_id

ID string

Yes

destination_version

Current integer version

Yes

acknowledge_destination_access

boolean; must be true

Yes

Result: page metadata and mutationOutcome.

Requires native space administration and the expected trashed version. Title, child, state and destination conflicts are refused. Destination ancestry determines inherited access.

JSON
{
  "space_key": "DEV",
  "page_id": "123456",
  "version": 7,
  "destination_page_id": "789",
  "destination_version": 3,
  "acknowledge_destination_access": true
}

Changes during an operation

Permission, hierarchy or content changes can stop copying, moving or restoration. Saved items may remain. Inspect current native state and the returned outcome before retrying; inaccessible records may be withheld.

For connection and OAuth details, see MCP connection reference.