| name | confluence |
| description | Read Confluence pages and maintain Confluence tracking — search, fetch, record page IDs in sync registry and article frontmatter, and transform page content into resource facts. Load when fetching a Confluence page, enriching a resource from Confluence, or updating meta/confluence-sync.json. |
| updated | "2026-05-07T00:00:00.000Z" |
| related | ["resources","query"] |
Confluence read/write CLI authority and Confluence → resource transformation rules.
> Read-only by default. NEVER create/edit/delete pages without explicit user approval. Distill durable facts into resource articles; record provenance in front matter `confluence:` and monitoring in `meta/confluence-sync.json`. For resource article format, load `resources` skill. For multi-source retrieval context (Confluence = Tier 3), load `query` skill.
<write_policy>
NEVER create, edit, or delete Confluence pages without explicit user approval.
Default: read-only. Before any write: ask "Should I write this to Confluence?" Proceed only on explicit yes.
CONFLUENCE_READ_ONLY=true enforces at CLI level.
</write_policy>
1. Check article's `confluence:` front matter for existing page IDs. If cited and content sufficient, DO NOT re-fetch.
2. Check `meta/confluence-sync.json` for last `synced` date. If recent, local content may be current.
3. If neither sufficient, fetch.
```bash
confluence read PAGE_ID # Preferred — unambiguous
confluence read PAGE_ID --format markdown # Markdown
confluence info PAGE_ID # Metadata only
```
| Goal | Command |
|------|---------|
| Keyword search | `confluence search "keyword" --limit 10` |
| Title in space | `confluence find "Title" --space SPACEKEY` |
| Browse children | `confluence children ROOT_PAGE_ID --recursive --format tree` |
| Discover spaces | `confluence spaces` |
USE ≥2 strategies before concluding page does not exist.
A. Article confluence: front matter (provenance):
confluence: [123456789]
Records which pages contributed facts. Many-to-many.
B. meta/confluence-sync.json (monitoring):
"PAGE_ID": {
"title": "Page Title",
"space": "SPACE_KEY",
"synced": "YYYY-MM-DD",
"resources": ["resources/domain/article.md"]
}
| Field | Content |
|---|
title | Exact Confluence title |
space | Space key (e.g. ENG) |
synced | Last ingest date YYYY-MM-DD, or null |
resources | Optional hint list of article paths |
Only register pages worth monitoring. Discovery scripts:
node {{AGENT_DIR}}/scripts/confluence-missing.js — untracked pages in spacenode {{AGENT_DIR}}/scripts/confluence-updates.js YYYY-MM-DD — changed since date
**Extract:** durable facts only — decisions, process definitions, architecture, ownership, team structure.
**Discard:** meeting logistics, status updates, changelogs, formatting boilerplate.
**Condense:** 5-page Confluence doc → ~30-line resource article. Use tables/bullets. Quote numbers, dates, names.
**Split:** one page → multiple articles when distinct topics. Cross-link them.
**Merge:** multiple pages → one article when describing one topic. Track all IDs in `confluence:`.
## Sources section format:
- [Page Title](<confluence-url>/spaces/SPACE/pages/PAGE_ID) — reason this page was used
Ensure `CONFLUENCE_READ_ONLY=false` or unset.
- Create:
confluence create "Title" SPACEKEY --file content.md --format markdown
- Create child:
confluence create-child "Title" PARENT_PAGE_ID --file content.md --format markdown
- Update:
confluence update PAGE_ID --file content.md --format markdown — show full new content first.
- Delete:
confluence delete PAGE_ID --yes — confirm ID and title. Irreversible.
- Move:
confluence move PAGE_ID NEW_PARENT_ID — same space only.
After any write: update meta/confluence-sync.json.
- Read-only by default. Ask before any write.
- Check frontmatter and `sync.json` first. DO NOT re-fetch sufficient content.
- Distill, never copy. Extract durable facts.
- Record provenance: add page IDs to `confluence:` front matter when extracting facts.
- `confluence read` accepts only page IDs or URLs with `pageId` query param — NOT display/pretty URLs.
<self_review>
<output_rules>Output in English. Preserve verbatim CLI commands, page IDs, and JSON structure.</output_rules>
