Spec do Epico A - adiciona Camada 3 (memoria semantica episodica do contato)
ao Captain AI, mantendo as 3 camadas existentes inalteradas.
Decisoes fechadas no brainstorming:
- Extracao ao resolver conversa OU silencio > 30min (100% automatico)
- Validacao: evidence obrigatoria, confidence >= 0.5 (alternativas B/C/D
documentadas como fallback)
- Scope global no recall, atribuicao por source_unit_id pra relatorios
- 9 tipos iniciais, limite 5 fatos/conversa, 50 ativos/contato
- TTL por tipo + supersedencia automatica por contradicao
- LGPD soft-30d -> hard-delete via cron
- 2 feature flags independentes, default OFF
- Epico B (LangGraph/inteligencia) sera spec separado pos-producao
Custo estimado: ~R$ 47/mes no grupo todo.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(whatsapp): allow converting inbox between WhatsApp providers
Adds a Convert flow to switch a WhatsApp inbox between the four
supported providers (default/360dialog, whatsapp_cloud, baileys, zapi)
without losing conversations, agents, or history.
- Channel::Whatsapp#convert_provider! runs inside a transaction:
disconnects the old provider, clears provider_connection and
message_templates, assigns the new provider/config, and triggers
webhook setup plus template resync on the new service.
- New POST /api/v1/accounts/:id/inboxes/:id/convert_provider endpoint
guarded by InboxPolicy#convert_provider? (admin only).
- UI adds a Convert button on the inbox Settings page with a
type-to-confirm ConvertInboxModal that lists the effects before
redirecting to a dedicated route reusing the WhatsApp provider
wizard in convert mode (phone number locked, current provider
hidden from the picker).
* chore(whatsapp): polish convert UI colors and expand specs
- Settings: use slate for the Convert trigger and ruby for the modal
confirm to mirror the delete gate instead of the less conventional
amber variant.
- Drop the redundant "current provider is hidden from the list"
sentence from the convert wizard description.
- Add specs for the post-conversion webhook setup path (triggered and
skipped branches) and the sync_templates error-rescue behaviour.
* fix: address CodeRabbit review on convert-provider flow
- Whitelist provider_config keys in the convert endpoint via permit
rather than permit!, and default to an empty hash when omitted so
the request no longer crashes.
- Pre-validate the new provider config before disconnecting the old
session so a bad target config no longer terminates the existing
provider; also keep the disconnect bound to the old provider_url.
- Guard ConvertInboxModal's submit handler so pressing Enter cannot
bypass the type-to-confirm gate, and migrate it to <script setup>.
- Reject invalid ?provider= query values in convert mode so hidden
providers (Twilio, the current provider) cannot be reached via URL.
- Await the inbox fetch in InboxConvert before running the route guard
so directly opening the route for a non-WhatsApp inbox redirects.
- Remove the unreachable second CloudWhatsapp branch in Whatsapp.vue.
* fix: address second CodeRabbit round on convert-provider flow
- Unify provider picker validation so create mode also rejects
unknown ?provider= values, with a single helper that accepts
available providers plus the whatsapp_manual fallback.
- Simplify the pre-validation rollback in convert_provider!: the
errors snapshot/merge dance was redundant because assign_attributes
does not clear errors.
- Follow the repo convention of asserting on error.class.name so the
rollback spec stays stable under reloading/parallel environments.
- Strengthen the controller success spec with provider_connection and
message_templates cleanup invariants, and set Content-Type on the
templates stub so HTTParty parses the empty data array correctly.
* fix: address third CodeRabbit round on convert-provider flow
- Add 360Dialog entry to the Whatsapp provider catalog, keep it hidden
from the create picker (preserving the existing fork behavior) but
expose it in the convert picker where it is a valid target. Restore
URL reachability for ?provider=360dialog in create mode.
- Scope the WHATSAPP_MANUAL allowance to create mode only: the manual
fallback flow is not reachable in convert mode.
- Redirect to the inboxes list in InboxConvert when the inbox is still
absent after the store fetch, so the page no longer stays blank.
- Use an explicit allowlist of WhatsApp providers to gate the Convert
button instead of negating Twilio, so adding a new WhatsApp channel
type will not silently expose the flow.
- Bind the disabled provider display field with :value instead of
v-model, since the underlying computed is getter-only.
- Add Content-Type: application/json to the templates stub in the
model spec so HTTParty parses the empty data array.
* fix: address fourth CodeRabbit round on convert-provider flow
- Reject no-op conversions that target the same provider as the one
already configured, so the endpoint no longer wipes provider
connection and message templates on a request that changes nothing.
- Call the provider service's disconnect directly so failures abort
the conversion instead of being silently swallowed; otherwise the
old external session could remain live while the inbox flips to
the new provider.
- Cover both behaviors with specs.
* fix: address fifth CodeRabbit round on convert-provider flow
- Reset the Vuelidate state when closing ConvertInboxModal so reopening
the gate does not surface stale validation errors.
- Call teardown_webhooks before converting away from whatsapp_cloud so
the Meta webhook subscription is removed for embedded_signup channels,
mirroring the destroy-time cleanup (manual-setup channels keep the
existing no-op behavior). Swallow teardown failures so a flaky Meta
call does not abort the swap.
- Switch the rollback specs to compare message_templates counts instead
of the boolean be_present matcher so they remain meaningful if the
fixture happens to have an empty templates list.
* fix: address sixth CodeRabbit round on convert-provider flow
- Derive the convert header's current-provider label from the shared
PROVIDER_CATALOG so the picker and header stay in sync.
- Assert the full Cloud provider_config payload and the absence of the
Baileys-only provider_url key on both the controller success spec
and the model atomic-swap spec.
- In the sync-error spec, reload and assert that the record was
actually flipped to the new provider before the sync rescue fires,
so the test can't pass on a pre-save failure.
* test: pin 422 error payload on convert_provider negative paths
The unsupported-conversion and invalid-config specs only checked the
status code, so they would have stayed green if the 422 started coming
from a different branch. Pin the response body so each example actually
covers the failure case it names.
* fix(baileys): save custom host as provider_url, not url
The Baileys form was writing the custom endpoint to
provider_config['url'] while the backend reads
provider_config['provider_url']. That silently broke the custom-host
feature for newly created or converted Baileys inboxes: they always
fell back to BAILEYS_PROVIDER_DEFAULT_URL. Align the key on both ends.
* fix(whatsapp): skip second validation pass in convert_provider!
The transaction's save! was re-running validate_provider_config after
the old provider's session had already been disconnected, so a transient
Graph API failure on the second check could roll back the swap while
leaving the external session terminated — the exact inconsistency the
pre-flight valid? was meant to rule out.
Capture the validated provider_config snapshot after valid? (so fields
populated by before_validation callbacks like webhook_verify_token are
preserved) and switch the final persist to save!(validate: false) so the
earlier check stays authoritative.
* fix: normalize provider-conversion failures and pass accountId
- The convert_provider action only rescued ActiveRecord::RecordInvalid,
so disconnect/teardown failures bubbled up as 500 with no stable
payload. Catch StandardError, log the class + message, and return a
422 with a generic user-facing message so the dashboard can surface
the error consistently.
- Nested settings routes live under /accounts/:accountId, so the
router push from Settings.vue must include accountId alongside
inboxId. Mirrors how sibling pages navigate to settings_inbox_show.
* fix: report missing :provider as 400 and sync modal v-model
- The generic rescue StandardError on convert_provider was masking
ActionController::ParameterMissing behind a misleading
provider-conversion error message. Catch it explicitly before the
generic rescue and return 400 with the parameter-missing message.
- ConvertInboxModal's closeModal now drives localShow to false so
parents using v-model:show stay in sync on every close path,
not only when the explicit onClose listener flips the flag.
* fix(whatsapp): serialize concurrent convert_provider calls with_lock
Without a per-record lock, two admin requests against the same inbox
could both pass the pre-flight validation, race the disconnect/save,
and then run setup_webhooks/sync_templates in arbitrary order, leaving
the persisted provider out of sync with the external configuration.
Wrap the whole convert flow in with_lock so the loser blocks until the
winner commits; the subsequent no-op guard then rejects a second
conversion request targeting the provider the first one just set.
* test: harden convert_provider policy + controller failure specs
- Pass accountId explicitly in InboxConvert redirects so the route
navigation mirrors how Settings.vue reaches settings_inbox_convert.
- Add a spec that assigns the agent to the inbox and still expects 401,
so a future regression in InboxPolicy#convert_provider? can no longer
slip past on the show policy alone.
- Add a spec that stubs convert_provider! to raise StandardError and
asserts the controller's generic-failure 422 payload, pinning the
dashboard contract for provider-side failures.
* test: pin convert_provider success response payload
Parse the rendered body and assert provider + provider_config so the
spec catches regressions where the DB is updated correctly but the
serialized response drifts (dashboard store commits response.data).
* fix(whatsapp): reset teardown guard after pre-conversion webhook cleanup
teardown_webhooks memoizes @webhook_teardown_initiated = true to prevent
double execution during destroy. Calling it from convert_provider!
leaves that flag set, so a subsequent destroy! or follow-up conversion
on the same instance would skip webhook removal silently. Reset the
flag in an ensure block so the destroy-time guard stays scoped to
destroy only.
* fix: include accountId in post-conversion redirect params
* test: pin same-provider convert returns 422
* fix(whatsapp): reset template columns when post-conversion sync fails
* fix(convert): enforce provider allowlist in InboxConvert route guard
* test: broaden Cloud templates stub to match account-scoped path
* test(whatsapp): cover cloud to baileys conversion branch
Problema observado: Daniela chamou generate_pix com arguments vazios apos
cliente informar "27/4". Tool retornou missing_fields=[check_in, amount] e
LLM caiu no fallback silenciosamente.
Correcoes:
- DDMMYYYY_REGEX agora aceita "DD/MM" sem ano (assume ano corrente, empurra
pro proximo ano se a data ja passou)
- parse_date_without_year com fallback explicito
- Instruction da scenario Daniela_Reservas (DB, scenario_id=2) atualizada
para listar todos os 4 parametros obrigatorios de generate_pix e
distinguir requires_input (erro do LLM) de success=false (erro tecnico)
Backup da instruction anterior: /tmp/daniela_instruction_backup_20260418.txt
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Upstream migration 20260324102005_repurpose_response_bot_flag_for_custom_tools
calls InstallationConfig.value before the normalization migration
(20260418020000) runs. On instances whose serialized_value column holds
native jsonb hashes, that read raises "TypeError: no implicit conversion of
Hash into String" and aborts the whole migration chain, leaving db:migrate
stuck and production unable to deploy.
Restore SerializedValueCoder.load that tolerates either shape. dump is
updated to always emit YAML strings so new writes converge on the upstream
format; the 20260418020000 migration still normalizes the backlog.
Some rows in installation_configs.serialized_value were written as native
jsonb objects by older code paths, while the YAML coder expects JSON-encoded
YAML strings. Reading the object-shaped rows raised "TypeError: no implicit
conversion of Hash into String" in production after the upstream 4.13.0 merge.
Convert every object-shaped row to the YAML-string shape the coder produces,
so the stock serialize :serialized_value, coder: YAML, ... works for all
rows without needing a custom coder.
Inboxes without portal_id were crashing with NoMethodError on save,
blocking landing host creation via UI for any inbox without a portal.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Devise emails sent from unauthenticated flows (password reset,
unlock, confirmation re-send) were always rendered in the default
locale because Current.account is nil in that context. Use the
user's first account as a fallback so the email respects the
account locale configured by the user.
CHANNEL_SERVICES held frozen references to service class objects, captured
when SendReplyJob was first autoloaded. In test env (cache_classes=false),
Zeitwerk reloads triggered between specs (notably by request specs)
replaced the constants with new class objects, leaving the hash pointing
to the stale ones. RSpec stubs applied to the current constant were then
bypassed when the job called service_class.new(...) through the stale
reference, causing flaky CI failures in spec/jobs/send_reply_job_spec.rb
when sharded together with V2::ReportBuilder + Captain::Preferences specs.
Storing class names as strings and resolving via constantize per call
fixes this and is the standard Rails autoload-safe pattern.
The Rails schema dumper can't capture CREATE FUNCTION statements, so every
db:schema:dump silently drops the execute <<~SQL block that defines
f_unaccent, breaking schema loading downstream (the functional GIN
indexes reference the missing function).
Mirror the existing schema-load enhance hook with a post-dump task that
re-injects the block between the last enable_extension and the first
create_table. Idempotent: skips when the block is already present.
Also bundle the merge-upstream skill that documents the recurring merge
patterns for this fork.
users.message_signature is nullable, so currentUser.message_signature can
arrive as null for accounts without a signature set. Vue prop defaults
only kick in for undefined, so the null passed through v-model to the
Editor, which called MarkdownIt.parse(null) and threw 'Input data should
be a String', breaking the profile settings page.
Introduce a `Last Responding Agent` options to assign_agents action in
automations to cover the following use cases.
- Assign conversations to first responding agent : ( automation message
created at , if assignee is nil, assign last responding agent )
- Ensure conversations are not resolved with out an assignee : (
automation conversation resolved at : if assignee is nil, assign last
responding agent )
and potential other cases.
fixes: #1592
## Summary
The `assignee_name` and `user_name` variables are swapped in the Chinese
(zh/zh_CN) locale files for conversation assignment activity messages,
causing the rendered text to show the wrong person as the assignee.
### Before (incorrect)
| Template | English (correct) | Chinese (incorrect) |
|---|---|---|
| `assignee.assigned` | Assigned to **AgentA** by **Admin** | 由
**AgentA** 分配给 **Admin** |
| `team.assigned` | Assigned to **TeamX** by **Admin** | 由 **TeamX** 分配给
**Admin** |
| `team.assigned_with_assignee` | Assigned to **AgentA** via **TeamX**
by **Admin** | 由 **AgentA** 分配给 **TeamX** 团队的 **Admin** |
The Chinese text reads as if the conversation was assigned **to Admin**
(the API caller), when it was actually assigned **to AgentA**.
### After (correct)
| Template | Chinese (fixed) |
|---|---|
| `assignee.assigned` | 由 **Admin** 分配给 **AgentA** |
| `team.assigned` | 由 **Admin** 分配给 **TeamX** |
| `team.assigned_with_assignee` | 由 **Admin** 通过 **TeamX** 团队分配给
**AgentA** |
Now correctly matches the English template semantics.
## Files Changed
- `config/locales/zh_CN.yml` — 3 lines
- `config/locales/zh.yml` — 3 lines
## How to Verify
1. Set locale to `zh_CN`
2. Have an admin assign a conversation to an agent
3. Check the activity message in the conversation — the assignee name
should appear after "分配给", not before it
Co-authored-by: Sojan Jose <sojan@pepalo.com>
This updates macros and automations so agents can explicitly remove
assigned agents or teams, while keeping the existing `Assign -> None`
flow working for backward compatibility.
Fixes: #7551Closes: #7551
## Why
The original macro change exposed unassignment only through `Assign ->
None`, which made macros behave differently from automations and left
the explicit remove actions inconsistent across the product. This keeps
the lower-risk compatibility path and adds the explicit remove actions
requested in review.
## What this change does
- Adds `Remove Assigned Agent` and `Remove Assigned Team` as explicit
actions in macros.
- Adds the same explicit remove actions in automations.
- Keeps `Assign Agent -> None` and `Assign Team -> None` working for
existing behavior and stored payloads.
- Preserves backward compatibility for existing macro and automation
execution payloads.
- Downmerges the latest `develop` and resolves the conflicts while
keeping both the new remove actions and current `develop` behavior.
## Validation
- Verified both remove actions are available and selectable in the macro
editor.
- Verified both remove actions are available and selectable in the
automation builder.
- Applied a disposable macro with `Remove Assigned Agent` and `Remove
Assigned Team` on a real conversation and confirmed both fields were
cleared.
- Applied a disposable macro with `Assign Agent -> None` and `Assign
Team -> None` on a real conversation and confirmed both fields were
still cleared.
# Pull Request Template
## Description
This PR adds support for resizing the reply editor up to nearly half the
screen height. It also deprecates the old modal-based pop-out reply box,
clicking the same button now expands the editor inline. Users can adjust
the height using the slider or the expand button.
## Type of change
- [x] New feature (non-breaking change which adds functionality)
## How Has This Been Tested?
### Loom video
https://www.loom.com/share/be27e1c06d19475ab404289710b3b0da
## Checklist:
- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [x] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
---------
Co-authored-by: Pranav <pranav@chatwoot.com>
Update BulkSelectBar to compute selection state (indeterminate/all) from
visible item IDs and only toggle selection for visible items. Preserve
existing selection for off-screen items when toggling, and guard against
empty visibility. Add detection/rendering for an optional
secondary-actions slot and adjust layout/divider. Also fix
ContactsBulkActionBar selection logic to determine "all selected" by
verifying every visible ID is in the selection. These changes ensure
correct select-all behavior with filtered/visible lists and support
additional UI actions.
https://github.com/user-attachments/assets/d06b78d1-a64a-4c0c-a82a-f870140236c7
# Pull Request Template
## Description
Please include a summary of the change and issue(s) fixed. Also, mention
relevant motivation, context, and any dependencies that this change
requires.
Fixes # (issue)
## Type of change
Please delete options that are not relevant.
- [ ] Bug fix (non-breaking change which fixes an issue)
- [ ] New feature (non-breaking change which adds functionality)
- [ ] Breaking change (fix or feature that would cause existing
functionality not to work as expected)
- [ ] This change requires a documentation update
## How Has This Been Tested?
Please describe the tests that you ran to verify your changes. Provide
instructions so we can reproduce. Please also list any relevant details
for your test configuration.
## Checklist:
- [ ] My code follows the style guidelines of this project
- [ ] I have performed a self-review of my code
- [ ] I have commented on my code, particularly in hard-to-understand
areas
- [ ] I have made corresponding changes to the documentation
- [ ] My changes generate no new warnings
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [ ] New and existing unit tests pass locally with my changes
- [ ] Any dependent changes have been merged and published in downstream
modules
---------
Co-authored-by: Sivin Varghese <64252451+iamsivin@users.noreply.github.com>
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
Co-authored-by: iamsivin <iamsivin@gmail.com>
