| name | notion-workflow-spovishun |
| description | Project-specific Notion workflow for the Spovishun project. Always use this skill at the start of any task that involves the SpovishunTelegramBotV2 project and Notion — even if it's just reading a page, creating a task, or updating docs. Triggers on any mention of "Spovishun", "SpovishunBot", "TelegramBotV2", "spovishun bot", or any related variation. |
Notion Workflow — Spovishun Project
Workflow
- Fetch the targeted section of CLAUDE.md for the current operation (see commands below).
- Identify the operation type and pick the matching row in the Decision Table.
- If IDs are needed, read
references/data-ids.md — do not guess or recall IDs from memory.
- Invoke the routed skill to complete the operation.
CLAUDE.md Fetch Commands
Fetch silently before responding — no need to announce it:
node scripts/notion/get-claude-md.js --section commands # architecture / commands
node scripts/notion/get-claude-md.js --section testing # testing conventions
node scripts/notion/get-claude-md.js --section architecture # source structure / layers
node scripts/notion/get-claude-md.js # full read — only when overview needed
Use targeted --section reads to load only the relevant part and save tokens.
REST-first hot path
The start-task flow (notion-task-inject.js hook + scripts/notion/*.js) is 100% Notion REST via scripts/notion/lib/notion-http.js. Task content, status updates, and CLAUDE.md reads all happen through these scripts — no MCP is involved. The hook writes the result to .dev-context/{branch}_prd/ and injects it via additionalContext (no tool-call overhead for Claude).
Do NOT add MCP calls to this path. MCP is for operations the scripts cannot cover.
MCP vs REST decision matrix
| Operation | Tool |
|---|
| Read task / board / CLAUDE.md (hot path) | scripts/notion/*.js (REST) |
| Update task status (hot path) | scripts/notion/update-status.js or hook PATCH |
| Free-form semantic search across Notion | MCP notion-search |
| Create / update arbitrary page content | MCP via notion-page-builder skill |
| Create / query databases | MCP via notion-database-manager skill |
| Bulk doc sync | MCP via update-doc-full |
Decision Table
| Operation | Skill |
|---|
| Task CRUD on Spovishun board | notion-spovishun-task-manager |
| Generic Kanban board operations | notion-task-board-manager |
| Create / update Notion pages | notion-page-builder |
| Search / read existing content | notion-content-reader |
| Create / query databases | notion-database-manager |
| Migrate external content | notion-data-migrator |
| Workspace structure / moves | notion-workspace-organizer |
| Workspace / category / collection IDs | notion-navigator |
| Board IDs, CLAUDE.md page ID, data source ID | references/data-ids.md |
Do NOT
- Do NOT load
references/data-ids.md unless an ID is actually needed — most operations go through notion-navigator or the dedicated skill.
- Do NOT duplicate operations handled by a routed skill — delegate and do not re-implement.
- Do NOT hardcode IDs from memory; always look them up from
references/data-ids.md or notion-navigator.
Error Handling
- If no row in the Decision Table matches, ask the user to clarify the operation type.
- If a reference file is missing, stop and report the exact path.
- If
node scripts/notion/get-claude-md.js fails, check that NOTION_API_TOKEN is set in the environment.
Related Skills
notion-navigator — full workspace map with all category page IDs and collection IDsnotion-spovishun-task-manager — task CRUD, status updates, board queriesnotion-page-builder — creating and updating Notion page content via MCP
Example Invocation
- User: "Read the current task" → fetch CLAUDE.md commands section, load
notion-spovishun-task-manager.
- User: "Update the Architecture doc page" → load
notion-navigator for the correct parent ID, then notion-page-builder.
