shared/input-classifiers/confluence-url

Source: shared/input-classifiers/confluence-url.md

input classifier: Confluence URL

Pattern

1https?://<site>.atlassian.net/wiki/spaces/<SPACE>/pages/<pageId>/<title-slug>2https?://<site>.atlassian.net/wiki/spaces/<SPACE>/pages/<pageId>

Fetch via

adk-mcp-atlassian → getConfluencePage(pageId) (returns content in storage format / ADF / Markdown depending on flag).

Extract into context.md

1### [confluence] <title>2url: <full URL>3space: <SPACE>4page id: <pageId>5last edited: <date> by <author>6labels: [...]7summary (first 80 words of body, plain text): <quote>8key sections (h2/h3):9  - <section title> — <one-line summary>10linked pages: [...]11attachments: <count, names>

Hints

• /adk-implement: if the page is a TDD/spec, parse it into acceptance criteria.
• /adk-document --type adr | --type runbook: if updating, fetch the current version first.
• /adk-sync --read: this is the primary read path.

Anti-patterns

• Don’t paste the entire page body into context. Quote sparingly; link out.
• If the page has attachments (images, especially), note their existence but don’t fetch unless asked.