---

name: confluence-integration description: "Manage Confluence pages - fetch, search, create, update. Requires Confluence credentials in .envrc." autoInvoke: true priority: medium triggers:

• "Confluence URL"
• "confluence page"
• "wiki page"
• "documentation request" allowed-tools: Read, Bash

---

Aura Frog Confluence Integration

Priority: MEDIUM - Use when Confluence page/documentation needed

---

When to Use

AUTO-DETECT when:

• Confluence URL shared: https://company.atlassian.net/wiki/...
• Documentation request with Confluence mention
• Phase 8 (Documentation) - for publishing
• Phase 9 (Share) - for updating team docs
• User mentions "confluence page", "wiki", "documentation"

---

Integration Process

1. Detect Confluence Reference

// Confluence URL pattern
const urlMatch = message.match(/https:\/\/[^\/]+\.atlassian\.net\/wiki\/.*pageId=(\d+)/)
const pageId = urlMatch ? urlMatch[1] : null

// Or detect keywords
const keywords = ['confluence', 'wiki page', 'documentation page', 'confluence page']

2. Check for Existing Log Data (PRIORITY)

⚠️ IMPORTANT: Always check for cached data first before fetching!

Log file locations (check in order):

.claude/logs/confluence/page-{PAGE_ID}.html    ← Page content
.claude/logs/confluence/page-{PAGE_ID}.json    ← Full metadata
logs/confluence/page-{PAGE_ID}.html            ← Alternative
logs/confluence/page-{PAGE_ID}.json            ← Alternative

Decision logic:

IF log file exists AND user did NOT request "fetch", "refresh", "update":
  → READ from log file using Read tool
  → Show cached data with note: "📋 Using cached data"

IF log file does NOT exist OR user requests fresh data:
  → FETCH using bash script

3. Operations

Fetch Page (Read-Only)

# By page ID
bash scripts/confluence-operations.sh fetch 123456

# By title (requires space key)
bash scripts/confluence-operations.sh fetch "API Documentation" DEV

Output:

• logs/confluence/page-{ID}.html - Content
• logs/confluence/page-{ID}.json - Full metadata

Search Pages

# Search all spaces
bash scripts/confluence-operations.sh search "deployment guide"

# Search specific space
bash scripts/confluence-operations.sh search "release notes" PROJ 20

Output:

• logs/confluence/search-results-{timestamp}.json

Create Page (WITH CONFIRMATION)

bash scripts/confluence-operations.sh create DEV "API Documentation" docs/api.md
bash scripts/confluence-operations.sh create PROJ "Sprint Report" report.md 123456

⚠️ REQUIRES user confirmation before executing

Update Page (WITH CONFIRMATION)

bash scripts/confluence-operations.sh update 123456 docs/updated-content.md
bash scripts/confluence-operations.sh update 123456 docs/api.md "New Title"

⚠️ REQUIRES user confirmation before executing

---

Setup

In .envrc:

export CONFLUENCE_URL="https://company.atlassian.net/wiki"
export CONFLUENCE_EMAIL="email@company.com"
export CONFLUENCE_API_TOKEN="your-token"

Get API Token:

1. Go to https://id.atlassian.com/manage-profile/security/api-tokens
2. Create new token
3. Add to .envrc

---

Phase Integration

Phase	Confluence Use
Phase 1	Search existing docs for context
Phase 2	Reference architecture docs
Phase 8	Create/update implementation docs
Phase 9	Publish summary, notify stakeholders

---

Safety Rules

ALWAYS require confirmation for:

• ✅ Creating new pages
• ✅ Updating existing pages
• ✅ Moving pages

READ operations are safe:

• ✅ Fetch page content
• ✅ Search pages

---

Confirmation Format

Before create/update operations:

⚠️ **CONFIRMATION REQUIRED: [Create/Update] Confluence Page**

**Space:** DEV (Development)
**Title:** "API Documentation"
**Page ID:** 123456 (if update)

**Action:** [Create new page / Update version X → Y]

Type "confirm" to proceed or "cancel" to skip

---

📚 Setup: docs/INTEGRATION_SETUP_GUIDE.md 📚 Agent: agents/confluence-operations.md

Remember: Confluence integration is optional. Workflow works without it.