Changelog

Add [QUERY] block support to Duncan's cloud chat, allowing dynamic SQL

execution against embedded datasets. When a note contains dataset-backed

tables, Duncan now receives full schema metadata (column names, types,

row count) and can emit [QUERY]SELECT ... FROM source:"Name"[/QUERY]

blocks. The system executes queries via IDatasetService.ExecuteSqlV2Async,

injects results into conversation history, and triggers a follow-up AI

call for analysis.

Safety: read-only validation (SELECT/WITH only), 100 row cap per query,

max 3 queries per response, cloud-only (local models excluded).

Cloud models use the user's own API key — they should speak freely.

The tier-based token ceilings (1024/4096/8192) and sentence-count

truncation were designed for local models that can't self-regulate.

Changes:

• CloudMaxTokensFor now returns a flat 16384 regardless of tier
• Sanitize() accepts isCloud flag: skips sentence truncation and

markdown stripping for cloud responses

• Local model path unchanged (still gets tight limits + cleanup)

Gemini 2.5 Pro tends to be verbose in its JSON responses, causing

truncation at 256 tokens which breaks JSON parsing. 512 gives enough

headroom for the memory extraction response without significant cost.

The note entity JSON nests blocks at content.content (PageDocument has

{ "type": "doc", "content": [...blocks...] }), not at content directly.

The previous fix checked content.ValueKind == Array which was Object

(the PageDocument), so it fell through to the else branch and returned

without finding any dataset references.

Now drills content → content.content to find the block array.

AppendEmbeddedDatasetContentAsync expected note content to be a JSON

string and scanned for "dataset_id" via text search. Content is actually

a JSON array of typed block objects, so the method bailed at the

type check and never found any dataset references.

Rewritten to:

• Parse content as a block array, iterating table/chart blocks
• Extract dataset_id (Data plugin datasets) and provider_plugin_id +

provider_query_key (cross-plugin queries from Tasks, Contacts, etc.)

• Query via IPluginDataSourceProvider using the correct provider for

each backing mode

• Fetch first 20 rows and inject columns + data into Duncan's context
• Legacy string fallback preserved for older content formats
• Capped at 3 embedded data sources per note to stay within token limits