Omniscio documentation
Browse all documentation
  1. Getting Started13
  2. Sessions & Agents115
  3. Inbox & Notifications59
  4. Projects & Tasks95
  5. Automation & Scheduling75
  6. Knowledge & Memory26
  7. AI Features60
  8. Integrations100
  9. Plugins & Marketplace33
  10. Cloud & Teams56
  11. Settings & Customization58
  12. Account & Billing28
  13. Troubleshooting84
  14. CLI & API Reference22
  15. Legal & Policies4
  16. Uncategorised22

PM Search Integration

PM items are discoverable through the global search alongside sessions, SMS, Slack, Telegram, RSS and webhook results, by both text search and semantic embedding similarity, with a "Use as context" action and session-linked deduplication.

PM items are discoverable through the global search (Ctrl+K) alongside sessions, SMS, Slack, Telegram, RSS, and webhook results -- via both text search and semantic embedding similarity.

What it is

PM items appear in global search results with a KanbanSquare icon, showing status, assignee, and due date in the snippet. Clicking navigates to the Mission Control virtual project. A "Use as context" action in the item drawer copies the item's full context (fields, subitems, updates) to clipboard for pasting into a session.

A PM_SUGGEST_ITEMS IPC channel returns semantically similar PM items for a text query.

When a session is linked to a PM item, the global search deduplicates: the standalone PM result is removed and the session result represents both.

Where to find it

Open the global search anywhere in Omniscio with Ctrl+K and type; PM items come back alongside the other result kinds, each labelled with its status, assignee and due date. Selecting one navigates to the Mission Control virtual project, and from the item drawer the "Use as context" action is where you copy an item's full context for pasting into a session.

How it behaves

Embeddings

Embeddings are stored in pm_item_embeddings as 384-dimensional float32 blobs (1536 bytes each). Source text combines the item name with all column value texts. The embedding model runs in a utilityProcess child.

Key behaviors:

  • Re-embedding is debounced (500ms) after item changes
  • After board sync, all synced items are scheduled for re-embedding
  • Backfill processes items without embeddings in batches of 50
  • Semantic search uses brute-force cosine similarity (capped at 10k embeddings, threshold 0.4)
  • All operations check getEmbeddingModelStatus() === 'ready' first

Settings

All PM search stages are gated on missionControlEnabled. The "Use as context" action only copies text -- it never spawns a session.

For agents

IPC channels and CLI routes

  • Global search (SEARCH_ALL) includes PM text-search and semantic stages
  • PM_SUGGEST_ITEMS -- semantic item suggestions
  • CLI: GET /pm/search?q=&boardId=&assignee=&status= for board-scoped and cross-board search with filters

The feature's obligations are pinned in pm-search-integration-contract.md under .claude/memory/contracts/.

Related

The results themselves come from the boards, so mission-control.md is the parent feature. global-search.md covers the search surface these results appear in, and semantic-search.md covers the embedding-based matching PM items share with the other content types.

Last verified 2026-09-23