Menu
Background AI drafting for Dataspheres AI pages
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
Local dev skill for the Dataspheres AI Content API. Wraps /api/v1/ REST endpoints for local-to-production content workflows. Use when the user wants to push pages, generate release notes from git log, list pages, or update content in a datasphere from their local machine.
Full newsletter lifecycle — create, configure all settings (frequency, personalization, AI model, web search, reply threading, plan mode wiring), manage subscribers, attach forms, draft and manage issues, preview personalized letters, enable private chat and email replies, and test in dev.
Drive the Dataspheres AI platform from Claude Code — read conversation history, post messages as the user (via API key), poll for ARI replies, read the Reality Engine debug log, update the plan and outcomes, and control orchestration flow. Use when you need Claude Code to interact with ARI or inspect/modify a running reality session.
Knowledge-graph tools for Dataspheres AI — build typed graphs, relate nodes with VISUAL or executable TASK edges, group into colored container bubbles, auto-detect article hero images, embed graphs in pages, run scheduled searches, and report.
Sequencers tools for Dataspheres AI
Manage Kanban tasks, plan modes, and project workflows in Dataspheres AI
| name | ai |
| description | Background AI drafting for Dataspheres AI pages |
| argument-hint | [action] [options] |
The AI Drafter generates page content asynchronously. Jobs run in the background and are attached to a specific page. The primary interface is background drafts — the streaming endpoint (POST /api/v2/ai/draft) is browser-only (SSE) and not suitable for MCP use.
result = draft_content(
content="Write a product update covering our Q2 launch",
context="<h1>Q2 Product Update</h1><p>We shipped X and Y this quarter.</p>",
page_id="<pageId>",
)
# → {"jobId": "job_abc123", "status": "PENDING"}
content — instructions / prompt for the AIcontext — existing page content (gives the AI surrounding material to work from)page_id — the page this draft is for (use get_page from the pages skill to find the ID)model_id — optional; defaults to claude-sonnet-4-6jobs = get_draft_jobs(page_id="<pageId>")
# → [{"jobId": "job_abc123", "status": "COMPLETED", "draftContent": "<h2>...</h2>"}]
Or fetch a single job:
job = get_draft_job(job_id="job_abc123")
# → {"jobId": ..., "status": "COMPLETED", "draftContent": "...", "error": null}
Status values: PENDING → PROCESSING → COMPLETED | FAILED
accept_draft(job_id="job_abc123") # applies the draft to the page
dismiss_draft(job_id="job_abc123") # discards it
| Tool | Method | Endpoint |
|---|---|---|
draft_content | POST | /api/v2/ai/draft/background |
get_draft_jobs | GET | /api/v2/ai/draft/jobs/:pageId |
get_draft_job | GET | /api/v2/ai/draft/job/:jobId |
accept_draft | POST | /api/v2/ai/draft/jobs/:jobId/accept |
dismiss_draft | POST | /api/v2/ai/draft/jobs/:jobId/dismiss |
POST /api/v2/ai/draft) is SSE and browser-only.research skill (assistant conversations) for general AI queries.Each draft_content call triggers an LLM completion. Confirm with the user before batch-drafting multiple pages.
| Error | Cause | Fix |
|---|---|---|
| 401 | Invalid API key | Re-run dai login |
| 404 on page_id | Page not found | Verify with get_page from the pages skill |
| 422 | Missing content or context | Both fields are required |