ana chat new: expose tool toggles (ontology / ontology-editing / auto-approve / web-search / sql / python) as flags

[bradfair]

Problem

ana chat new currently only accepts --connector and --title. The TextQL web UI exposes a richer set of tool toggles when creating a chat:

• Web Search
• JavaScript
• Bash
• Text to SQL
• Ontology
• Ontology Editing
• Auto Approve

There is no CLI path to enable these — every chat created with ana chat new defaults to whatever the server uses for unset fields. In practice that means the chat is created without ontology / ontology-editing / auto-approve, and any prompt that tries to drive ontology work via the CLI silently fails to do real ontology authoring.

This blocks the Phase 6 ontology-authoring step of the mock-data-forge skill and any other automation that wants to drive ontology builds from a script.

Repro

ana chat new --connector 7518 --title "Build ontology" --json
# → paradigm.options.universal only carries connectorIds + (default-on) sqlEnabled/pythonEnabled/webSearchEnabled.
# No ontologyEnabled / ontologyEditingEnabled / autoApproveEnabled even if you wanted them.

ana chat new --help
# Flags:
#   --connector value     connector id(s), comma-separated (required)
#   --endpoint string     override API endpoint URL
#   --json                emit JSON output
#   --profile string      config profile to use
#   --title string        optional chat summary/title
#   --token-file string   path to bearer-token file

Workaround

The underlying RPC already accepts the toggles — the CLI just doesn't surface them. Hitting CreateChat directly works:

ana api textql.rpc.public.chat.ChatService/CreateChat --data '{
  "summary": "Phase 6 ontology",
  "paradigm": {
    "type": "TYPE_UNIVERSAL", "version": 1,
    "options": {"universal": {
      "connectorIds": [7518],
      "pythonEnabled": true,
      "sqlEnabled": true,
      "webSearchEnabled": true,
      "ontologyEnabled": true,
      "ontologyEditingEnabled": true,
      "autoApproveEnabled": true
    }}
  }
}'

Response confirms the fields land:

{
  "chat": {
    "paradigm": {"options": {"universal": {
      "ontologyEnabled": true,
      "ontologyEditingEnabled": true,
      "autoApproveEnabled": true,
      ...
    }}}
  }
}

Subsequent ana chat send against that chat produces real ontologyEditorCell cells and authors objects in the connector's ontology.

Proposal

Add CLI flags on ana chat new that map 1:1 to the universal-paradigm toggles:

Flag	Effect
--ontology / --no-ontology	Set ontologyEnabled
--ontology-editing / --no-ontology-editing	Set ontologyEditingEnabled
--auto-approve / --no-auto-approve	Set autoApproveEnabled
--web-search / --no-web-search	Set webSearchEnabled
--sql / --no-sql	Set sqlEnabled
--python / --no-python	Set pythonEnabled
--javascript / --bash (?)	If those are exposed in the proto — currently the UI shows them as separate toggles

Defaults should match current server behavior so existing scripts don't change.

A grouped --tools <comma,sep,list> form would also work but is less discoverable — explicit per-toggle flags match the UI's mental model.