confluence

Confluence via confluence-cli (read, search, create, update, move, delete, attachments, comments). Shared npm CLI; per-workspace .env via SkillCred. Use when the user mentions Confluence, wiki pages,

SKILL.md

GitHub

TO COPY

confluence

Per-workspace registration slice. Credentials live in {SKILL_PATH}/.env.

Working directory

Placeholders changed by /meta-skills at copy time (NAME => {PLACEHOLDER}):

IS_GLOBAL => {IS_GLOBAL} TYPE_OF_AI_TOOLS => {TYPE_OF_AI_TOOLS} SKILL_PATH => {SKILL_PATH}

Bash / zsh:

export CURRENT_SKILL_DIRECTORY="{SKILL_PATH}"
eval "$(~/.meta-skills/.venv/bin/python ~/.meta-skills/skills/productivity/confluence/scripts/skill_env.py)"
confluence <subcommand> …

PowerShell (Windows):

$env:CURRENT_SKILL_DIRECTORY = "{SKILL_PATH}"
& "$HOME/.meta-skills/.venv/Scripts/python.exe" "$HOME/.meta-skills/skills/productivity/confluence/scripts/skill_env.py" --shell powershell | Out-String | Invoke-Expression
confluence <subcommand> …
END TO COPY

confluence

When to use

Use for Atlassian Confluence page/space work. Triggers: "Confluence page", "wiki", "search Confluence", "create page", /confluence_*.

Confirm with the user before write ops (create, update, move, delete, upload). Prefer CONFLUENCE_READ_ONLY=true for research-only agents.

Prerequisites

This skill documents confluence-cli (npm). Install once per machine:

npm install -g confluence-cli
confluence --version

Load credentials into the shell, then call confluence directly:

export CURRENT_SKILL_DIRECTORY="{SKILL_PATH}"
eval "$(~/.meta-skills/.venv/bin/python ~/.meta-skills/skills/productivity/confluence/scripts/skill_env.py)"
confluence read 123456789 --format markdown

PowerShell:

$env:CURRENT_SKILL_DIRECTORY = "{SKILL_PATH}"
~/.meta-skills/.venv/bin/python ~/.meta-skills/skills/productivity/confluence/scripts/skill_env.py --shell powershell | Invoke-Expression
confluence read 123456789 --format markdown

skill_env.py reads .env via SkillCred and prints export commands for the detected OS/shell.

Credentials — SkillCred .env

Variable Example
CONFLUENCE_DOMAIN company.atlassian.net
CONFLUENCE_API_PATH /wiki/rest/api (Cloud) or /rest/api (Server/DC)
CONFLUENCE_AUTH_TYPE basic or bearer
CONFLUENCE_EMAIL user@company.com (required when basic)
CONFLUENCE_API_TOKEN API token / PAT

Optional: CONFLUENCE_PROFILE, CONFLUENCE_READ_ONLY=true, CONFLUENCE_FORCE_CLOUD, CONFLUENCE_LINK_STYLE.

cp ~/.meta-skills/skills/productivity/confluence/.env.example "{SKILL_PATH}/.env"

Slash commands

Map /confluence_<action> → confluence <action> … (after skill_env.py exports are eval'd).

Read / search (confluence)

Slash CLI Description
/confluence_read confluence read <pageId|URL> [--format markdown|text|storage|html] Read page
/confluence_info confluence info <pageId> [--format json] Metadata
/confluence_find confluence find "Title" [--space KEY] Find by title
/confluence_search confluence search "query" [--limit N] [--cql] Search / CQL
/confluence_spaces confluence spaces List spaces
/confluence_children confluence children <pageId> [--recursive] [--format json|tree] Children
/confluence_comments confluence comments <pageId> [--format json] [--all] List comments
/confluence_attachments confluence attachments <pageId> [--download] [--dest DIR] List / download
/confluence_export confluence export <pageId> [--format markdown] [--dest DIR] Export

Write (confluence — confirm first; --yes on deletes)

Slash CLI Description
/confluence_create confluence create "Title" SPACE [--file path|--content …] [--format markdown] Create page
/confluence_create-child confluence create-child "Title" <parentId> [--file …] [--format markdown] Child page
/confluence_update confluence update <pageId> [--title …] [--file …] [--format markdown] Update
/confluence_move confluence move <pageId> <newParentId> [--title …] Move (same space)
/confluence_delete confluence delete <pageId> --yes Trash page
/confluence_edit confluence edit <pageId> [--output page.xml] Fetch storage XML
/confluence_comment confluence comment <pageId> --content "…" [--location footer] Add comment
/confluence_comment-delete confluence comment-delete <id> --yes Delete comment
/confluence_attachment-upload confluence attachment-upload <pageId> --file path [--replace] Upload
/confluence_attachment-delete confluence attachment-delete <pageId> <attId> --yes Delete attachment
/confluence_copy-tree confluence copy-tree <src> <parent> [title] [--dry-run] Copy page tree

Profiles / misc

Slash CLI Description
/confluence_profile_list confluence profile list List profiles
/confluence_profile_use confluence profile use <name> Switch profile
/confluence_convert confluence convert -i in.md -o out.xml --input-format markdown --output-format storage Offline convert
/confluence_api confluence api <endpoint> … Raw REST helper

Global: confluence --profile staging <command>.

Page IDs

Accept numeric IDs or Confluence URLs (?pageId=, /pages/<id>/…). Prefer ID or /pages/<id> over /display/<space>/<title>.

List all pages in a space

Do not paginate with confluence search … --start N (or content/search?…&start=N) on Confluence Cloud — the offset is often ignored → same page of results forever → infinite loops / huge delays.

Use the content list API and bump start while len(results) == limit:

confluence api "content?spaceKey=KEY&type=page&limit=50&start=0"
confluence api "content?spaceKey=KEY&type=page&limit=50&start=50"
# … until fewer than 50 results

If you must use content/search, follow _links.next (cursor), never a bare start.

Agent tips

  • Run eval "$(python scripts/skill_env.py)" once per shell before confluence ….
  • Destructive commands: always pass --yes.
  • Prefer --format markdown for agent text; --format json for parsing.
  • Read-only agents: CONFLUENCE_READ_ONLY=true in the workspace .env.
  • Folders have no body — use info, not read/edit.
  • Never use confluence search --start to crawl a whole space — see List all pages in a space.
  • Full reference: confluence --help / confluence-cli.

Notes

  • Never commit .env or API tokens.