confluence-publishlisted

# Publishing markdown to Confluence Turns a markdown file in a repository into a Confluence page, and keeps it that way. The file is the master; Confluence is the rendering. This skill **creates and updates only**. It has no delete verb anywhere - not for a page, not for an attachment. Removing something is a human job in the Confluence UI. ## Prerequisites Credentials are shared with `jira` and `confluence`. If `~/.dbhq/atlassian/config.json` does not exist, run `${CLAUDE_SKILL_DIR}/../_shared/scripts/atlassian-setup.sh`. ## The binding Each file carries its own page id in YAML frontmatter: ```yaml --- confluence: space: "98765" parent: "1234567" page_id: "8901234" --- ``` `page_id` is written back on the first publish. **Commit that change** - without it the next run creates a second page instead of updating the first. Get the space id from `${CLAUDE_SKILL_DIR}/../confluence/scripts/confluence-search.sh spaces`. It is the numeric id, not the key. ## Publishing ```bash ${CLAUDE_SKILL_DIR}/scripts/publish.sh docs/payment-correlation.md --dry-run ${CLAUDE_SKILL_DIR}/scripts/publish.sh docs/payment-correlation.md ``` **Always dry-run first.** It converts the whole body and reports what would happen. A `--dry-run` against a file with no `page_id` yet is pure local conversion and needs no credentials at all. A `--dry-run` against a file that already carries a `page_id` also makes one read-only call to the live page, to preview the round-trip gate below before yo