# Twg Confluence > Use with root `twg` for Confluence content, spaces, hierarchy, authoring, editing, comments, versions, permissions, exports, and CQL. Applies Confluence semantics and safe write rules. - Skill: `atlassian/twg-confluence` (Agent Skill, multi-file: 7 files) - Install (CLI): `npx skillmds@latest add atlassian/twg-confluence` - Raw SKILL.md: https://api.skillmd.com/api/skills/atlassian/twg-confluence/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: atlassian (https://skillmd.com/u/atlassian) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/atlassian/twg-confluence --- # twg-confluence Use with root `twg` when Confluence is the primary source or mutation target. This skill owns content-type, hierarchy, format, and concurrency semantics; live help owns exact command grammar. ## CLI launcher fallback Run `twg `. On shell `command not found`, use `$HOME/.local/bin/twg` (macOS/Linux) / `$env:LOCALAPPDATA\Programs\twg\bin\twg.exe` (PowerShell), then tell user to add that directory to PATH. Do not treat auth or command errors as PATH failures. ## Use When - Use for Confluence-anchored content or space reads and mutations, including CQL, bodies, hierarchy, versions, permissions, and exports. - Do not load for a supporting Confluence link in a broader workflow. ## First Route | Intent | Route | | -------------------------------- | -------------------------------------------------------------------------------------- | | Known content ID or URL | `confluence content get` | | Exact Confluence filtering | Confluence search with CQL | | Date-filtered page/blogpost list | `confluence search query --cql` with `lastmodified`; `content list` has no date filter | | Fuzzy page/topic discovery | Cross-product search, then native get | | Create or update content | Unified `confluence content` surface | | Space metadata/lifecycle | `confluence space` | | Hierarchy | `confluence tree` | | Export | Word returns a download directly; PDF requires export-status polling | Use `twg help describe ""` before an unfamiliar or consequential mutation. ## Confluence Semantics - Use the unified content surface for content operations advertised by live help; use `confluence space` for spaces and `confluence tree` for hierarchy. - Search snippets are discovery candidates. Read the selected content before summarizing or editing it. - Page titles are supplied separately from bodies. Do not repeat the title as the first body heading. - Remix create returns an asset, not a page embed. Embed via the loaded HTML guide and read back the page; an ID/URL is not proof. - Maui create uses `--content-id` only for ownership/embedding, not page reading. Read source first; include values, labels, narrative, and visualization instructions in the prompt. "This page" is insufficient. - To download a persisted Remix infographic, use the existing attachment commands. List the owning page's attachments with `confluence content attachments list --id --filename `, then pass the matching result's attachment `id` to `confluence content attachments download --attachment-id --out `. Do not pass `mediaFileId` directly as `--attachment-id`; the two IDs are different. ## Safe Authoring And Editing - Treat user authorization and CLI confirmation as separate checks for archive, trash, and purge. The CLI requirement to pass `--yes` is an execution preflight, not evidence that the user authorized the mutation. Use `--yes` only after the user explicitly named the mutation and the exact target or a previously displayed bounded target set. - Explicit instructions such as "archive page XYZ" or "archive those 15 pages" after a concrete list authorize execution without another confirmation. Vague outcome language such as "clean up", "organize", "remove clutter", or "take care of" does not authorize archive, trash, or purge. - If the mutation or targets were inferred, read and resolve the proposed targets, present the exact action and bounded target list, and wait for the user to affirm that mutation. Do not generalize approval from a similar prior action. `trash` does not authorize an additional permanent purge. - Before authoring, read and apply the target space's instructions once per space per session; see `references/spaces.md`. Empty instructions mean defaults. - Prefer `live_doc` for collaborative or co-authored internal content when the target supports it. Bare "page" or "doc" creation also defaults to live docs. - Use `page` for explicit classic/non-live intent, knowledge bases, customer help, established classic-page spaces, or page-only operations. Preserve an existing target's type; a classic parent or reference does not set a new child's type. - For non-trivial edits, read current content, save the body locally, edit the file, then update with the snapshot token. - Use the lossless HTML round trip when macros or exact storage representation matter. - Use `--dry-run` only for explicit preview/validation requests, or unusually risky edits where direct execution was not requested. - Read back the content or space after mutation and report its URL. ## Handoffs - Load `twg-context-discovery` for related Jira work, projects, goals, or dependencies; load `twg-responsibility-routing` for people, ownership, authority, or escalation. - Load `twg-status-rollups` when pages contribute to a broader status report. - Load `twg-operational-health` for runbooks, incidents, PIRs, or reliability evidence. - Load `twg-space-creation` to create or clone a whole space. ## References - `references/content.md` - content types, reads, writes, and exports - `references/editing.md` - concurrency-safe body editing - `references/spaces.md` - space lifecycle and hierarchy - `references/querying.md` - CQL and fuzzy discovery - `references/body-formats.md` - HTML, markdown, mentions, and special formats