ZSXQ Fetch#

Use this skill to query the local OpenClaw ZSXQ database for summarization and grounded Q&A.

When to use#

• The user wants summaries of recent ZSXQ posts by time window or by group.
• The user asks questions that should be answered from synced ZSXQ posts or downloaded documents.
• The user wants to inspect downloaded attachments or search document text.
• The user wants to check sync status or manually trigger a sync.

Preconditions#

• The API server is running. Check with:

bash

curl http://127.0.0.1:8000/health

If it is not running, start it from the project root:

bash

uvicorn app.main:app --host 0.0.0.0 --port 8000

The server automatically syncs all subscribed groups in the background every hour (configurable via SYNC_INTERVAL_SECONDS).

Core workflow#

1. Query the database directly — do NOT sync before every request. Data is kept fresh by the background auto-sync.
2. Pull only the smallest relevant slice of data.
3. Use /api/v1/topics, /api/v1/documents, and /api/v1/search_documents to answer questions.
4. Summarize with explicit scope: group, time range, or topic.

When to trigger a manual sync#

Only trigger a manual sync if:

• The user explicitly asks for the "latest" or "most recent" content AND
• /api/v1/sync_status shows the last sync was more than 1 hour ago.

Otherwise, trust the database content.

Useful endpoints#

• Check sync status (always check this first for time-sensitive requests):

bash

curl "http://127.0.0.1:8000/api/v1/sync_status"

• Trigger a manual sync (only if data is stale):

bash

curl -X POST "http://127.0.0.1:8000/api/v1/trigger_sync"

• Read recent topics:

bash

curl "http://127.0.0.1:8000/api/v1/topics?group_id=GROUP_ID&limit=50&offset=0"

• List groups:

bash

curl "http://127.0.0.1:8000/api/v1/groups/all"

• List downloaded documents:

bash

curl "http://127.0.0.1:8000/api/v1/documents?group_id=GROUP_ID&limit=50"

• Search document text:

bash

curl "http://127.0.0.1:8000/api/v1/search_documents?q=KEYWORD&group_id=GROUP_ID"

Summarization pattern#

• Pull recent topics for the requested group or across relevant groups.
• If documents may contain the answer, search them with keywords from the user question.
• Build the answer from retrieved records only. State when the answer is inferred from partial evidence.

Output guidance#

• For update summaries, group findings by topic or theme, not by raw post order.
• For question answering, cite the specific topic text or document match preview used.
• If the database has no relevant records, say so. Only recommend a manual sync if the last sync was more than 1 hour ago.