mirror of
https://github.com/Luchexx/wacrm.git
synced 2026-11-02 09:19:51 +00:00
efd72b131e
The automation builder exposed several developer-facing fields that made automations hard to configure from the UI (issue #216): - Keyword Match parsed/trimmed on every keystroke, so typing a comma-separated, multi-word list like "SEO, search engine optimization" was impossible — the comma and trailing space were stripped immediately. - Tag Added / Add Tag / Remove Tag asked for a raw tag_id. - Assign Conversation → Specific agent asked for a raw agent_id. - Send Template asked for a manually typed template name + language. Changes: - Keyword input keeps a local draft and only parses into the keywords array on blur / Enter, then re-displays the cleaned form. - New tag / agent / template pickers select existing resources by name while still storing ids / names internally, so the stored config shape (and the engine/validation) is unchanged. Resources load once at the builder root and are shared via context. - Tags & approved templates load from Supabase (RLS-scoped, matching the broadcast picker); members load from /api/account/members. - Every picker falls back to its original raw input when its list is empty (fresh account / older deployment), so automations stay authorable. Saved-but-missing ids are preserved when editing. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| app | ||
| components | ||
| hooks | ||
| lib | ||
| types | ||
| middleware.ts | ||