There was a fundamental difference in how resolution counts were
calculated between the agent summary and timeseries reports, causing
confusion for users when the numbers didn't match.
The agent summary report counted all `conversation_resolved` events
within a time period by querying the `reporting_events` table directly.
However, the timeseries report had an additional constraint that
required the conversation to currently be in resolved status
(`conversations.status = 1`). This meant that if an agent resolved a
conversation that was later reopened, the resolution action would be
counted in the summary but not in the timeseries.
This fix aligns both reports to count resolution events rather than
conversations in resolved state. When an agent resolves a conversation,
they should receive credit for that action regardless of what happens to
the conversation afterward. The same logic now applies to bot
resolutions as well.
The change removes the `conversations: { status: :resolved }` condition
from both `scope_for_resolutions_count` and
`scope_for_bot_resolutions_count` methods in CountReportBuilder, and
updates the corresponding test expectations to reflect that all
resolution events are counted.
## About timezone
When a timezone is specified via `timezone_offset` parameter, the
reporting system:
1. Converts timestamps to the target timezone before grouping
2. Groups data by local day/week/month boundaries in that timezone, but
the primary boundaries are sent by the frontend and used as-is
3. Returns timestamps representing midnight in the target timezone
This means the same events can appear in different day buckets depending
on the timezone used. For summary reports, it works fine, since the user
only needs the total count between two timestamps and the frontend sends
the timestamps adjusted for timezone.
## Testing Locally
Run the following command, this will erase all data for that account and
put in 1000 conversations over last 3 months, parameters of this can be
tweaked in `Seeders::Reports::ReportDataSeeder`
I'd suggest updating the values to generate data over 30 days, with
10000 conversations, it will take it's sweet time to run but then the
data will be really rich, great for testing.
```
ACCOUNT_ID=2 ENABLE_ACCOUNT_SEEDING=true bundle exec rake db:seed:reports_data
```
Pro Tip: Don't run the app when the seeder is active, we manually create
the reporting events anyway. So once done just use `redis-cli FLUSHALL`
to clear all sidekiq jobs. Will be easier on the system
Use the following scripts to test it
- https://gist.github.com/scmmishra/1263a922f5efd24df8e448a816a06257
- https://gist.github.com/scmmishra/ca0b861fa0139e2cccdb72526ea844b2
- https://gist.github.com/scmmishra/5fe73d1f48f35422fd1fd142ea3498f3
- https://gist.github.com/scmmishra/3b7b1f9e2ff149007170e5c329432f45
- https://gist.github.com/scmmishra/f245fa2f44cd973e5d60aac64f979162
---------
Co-authored-by: Sivin Varghese <64252451+iamsivin@users.noreply.github.com>
Co-authored-by: Pranav <pranav@chatwoot.com>
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
Currently, auto-assignment runs only during conversation creation or
update events. If no agents are online when new conversations arrive,
those conversations remain unassigned.
With this change, unassigned conversations will be automatically
assigned once agents become available. The job runs every 15 minutes and
uses a fair distribution threshold of 100 to prevent a large number of
conversations from being assigned to a single available agent. This will
be customizable later.
We were using UTM params on various branding urls which weren't
compliant to standard utm params and hence were ignored by analytics
tooling. this PR ensures that the params stays compliant with defined
standard
ref: https://en.wikipedia.org/wiki/UTM_parameters
## Changes
- updated utm tags on widget and survey urls
- added utm on helpcenter branding
---------
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
We now support searching within the actual message content, email
subject lines, and audio transcriptions. This enables a faster, more
accurate search experience going forward. Unlike the standard message
search, which is limited to the last 3 months, this search has no time
restrictions.
The search engine also accounts for small variations in queries. Minor
spelling mistakes, such as searching for slck instead of Slack, will
still return the correct results. It also ignores differences in accents
and diacritics, so searching for Deja vu will match content containing
Déjà vu.
We can also refine searches in the future by criteria such as:
- Searching within a specific inbox
- Filtering by sender or recipient
- Limiting to messages sent by an agent
Fixes https://github.com/chatwoot/chatwoot/issues/11656
Fixes https://github.com/chatwoot/chatwoot/issues/10669
Fixes https://github.com/chatwoot/chatwoot/issues/5910
---
Rake tasks to reindex all the messages.
```sh
bundle exec rake search:all
```
Rake task to reindex messages from one account only
```sh
bundle exec rake search:account ACCOUNT_ID=1
```
## Linear reference:
https://linear.app/chatwoot/issue/CW-4649/re-imagine-assignments
## Description
This PR introduces the foundation for Assignment V2 system by
implementing agent_capacity and their association with inboxes and
users.
## Type of change
- [ ] New feature (non-breaking change which adds functionality)
## How Has This Been Tested?
Test Coverage:
- Controller specs for assignment policies CRUD operations
- Enterprise-specific specs for balanced assignment order
- Model specs for community/enterprise separation
## 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: Pranav <pranav@chatwoot.com>
Added comprehensive Twilio WhatsApp content template support (Phase 1)
enabling text, media, and quick reply templates with proper parameter
conversion, sync capabilities.
**Template Types Supported**
- Basic Text Templates: Simple text with variables ({{1}}, {{2}})
- Media Templates: Image/Video/Document templates with text variables
- Quick Reply Templates: Interactive button templates
Front end changes is available via #12277
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Sivin Varghese <64252451+iamsivin@users.noreply.github.com>
There were customer reported issues with FAQs which were generated in a
different langauge than what they were expecting. The reason behind this
was that the language of the account was not considered in the prompt
provided. If the language of the content was say Spanish, and the
account locale was english. The output was not predicable. The output
depends on the model and the execution time.
This PR would update the prompt to behave consistently with the account
locale. Even though the content provided is in a different language, it
would generate FAQs in the account locale.
Changes:
- Updated the prompt to include a detailed expectation of the FAQs
quality along with the language
- Added specs for the services where the prompt generator is called.
Tested the prompt using Phoenix playground across GPT 5, GPT 4.1, GPT
4.0. The reasoning setting for GPT 5 needs to be low so that it doesn't
generate random questions like "What was this updated?"
This PR improves the voice call creation flow by simplifying
configuration and automating setup with Twilio APIs.
references: #11602 , #11481
## Key changes
- Removed the requirement for twiml_app_sid – provisioning is now
automated through APIs.
- Auto-configured webhook URLs for:
- Voice number callbacks
- Status callbacks
- twiML callbacks
- Disabled business hours, help center, and related options until voice
inbox is fully supported.
- Added a configuration tab in the voice inbox to display the required
Twilio URLs (to make verification easier in Twilio console).
## Test Cases
- Provisioning
- Create a new voice inbox → verify that Twilio app provisioning happens
automatically.
- Verify twiML callback
- Webhook configuration
- Check that both voice number callback and status callback URLs are
auto-populated in Twilio.
- Disabled features
- Confirm that business hours and help center options are
hidden/disabled for voice inbox.
- Configuration tab
- Open the voice inbox configuration tab → verify that the displayed
Twilio URLs match what’s set in Twilio.
## Linear reference:
https://linear.app/chatwoot/issue/CW-4649/re-imagine-assignments
## Description
This PR introduces the foundation for Assignment V2 system by
implementing assignment policies and their association with inboxes.
Assignment policies allow configuring how conversations are distributed
among agents, with support for different assignment orders (round_robin
in community, balanced in enterprise) and conversation prioritization
strategies
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?
Test Coverage:
- Controller specs for assignment policies CRUD operations
- Enterprise-specific specs for balanced assignment order
- Model specs for community/enterprise separation
Manual Testing:
1. Create assignment policy: POST
/api/v1/accounts/{id}/assignment_policies
2. List policies: GET /api/v1/accounts/{id}/assignment_policies
3. Assign policy to inbox: POST
/api/v1/accounts/{id}/assignment_policies/{id}/inboxes
4. View inbox policy: GET
/api/v1/accounts/{id}/inboxes/{id}/assignment_policy
5. Verify community edition ignores "balanced" assignment order
6. Verify enterprise edition supports both "round_robin" and "balanced"
- testing the flows after enterprise folder deletion
## 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: Pranav <pranavrajs@gmail.com>
Co-authored-by: Pranav <pranav@chatwoot.com>
## Reference
https://github.com/chatwoot/chatwoot/pull/12149#issuecomment-3178108388
## Description
setup_webhook was done before the save, and hence the meta webhook
validation might fail because of a race condition where the facebook
validation is done before we saving the entry to the database.
## Type of change
Please delete options that are not relevant.
- [ ] Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
- New inbox creation, webhook validation
- Existing inbox update, webhook validation
-
<img width="614" height="674" alt="image"
src="https://github.com/user-attachments/assets/be223945-deed-475a-82e5-3ae9c54a13fa"
/>
## 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: Muhsin Keloth <muhsinkeramam@gmail.com>
# Pull Request Template
## Description
Fixes
https://linear.app/chatwoot/issue/CW-5411/actionviewtemplateerror-activestorageunrepresentableerror
### Problem
API endpoints return 500 errors when conversations contain image
attachments that can't be processed by ActiveStorage (e.g., files with
non-ASCII filenames, corrupted images, or malicious XSS filenames).
Root Cause: Commit 6cab74139 removed the representable? safety check
from thumb_url, causing `ActiveStorage::UnrepresentableError` to bubble
up and crash the API when it encountered a malformed image file.
Fix: Rescue `thumb_url` method to catch UnrepresentableError and return
an empty string while logging problematic names for future debugging.
This ensures the messages/attachments api does not break due to a single
corrupted image file.
## Type of change
- [x] Bug fix (non-breaking change which fixes an issue)
## How Has This Been Tested?
- Added specs
## 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
- [x] I have made corresponding changes to the documentation
- [x] My changes generate no new warnings
- [x] 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
- [x] Any dependent changes have been merged and published in downstream
modules
WhatsApp templates without parameters (body-only templates like
notifications, confirmations) were failing to send with the error:
ArgumentError (Unknown legacy format: NilClass). This affected all
parameter-less templates across marketing messages, notifications, and
utility templates.
### Summary
Fixed flaky Instagram webhook specs that failed intermittently in cloud
environments due to shared let blocks creating conflicting inboxes. The
Instagram channel factory already creates an inbox automatically, but
tests were adding extra ones in shared contexts.
Moved channel/inbox creation to isolated test contexts to prevent race
conditions between Facebook Page and Instagram Direct tests.
### Testing
```
for i in {1..30}; do
echo "=== Run $i ==="
RAILS_ENV=test bundle exec rspec spec/jobs/webhooks/instagram_events_job_spec.rb --fail-fast || break
done
```
Previously, intermittent failures could be reproduced locally. With
these changes, tests achieve ~100% pass rate.
WhatsApp template message errors were not being properly handled because
the `@message instance` variable was only set in the `send_message`
method but not in `send_template`. When template sending failed, the
`handle_error` method couldn't update the message status due to the
missing @message reference, resulting in silent failures with no user
feedback.
This PR fixes flaky test failures in the Instagram webhook specs that
were caused by Redis mutex lock conflicts when
tests ran in parallel.
### The Problem:
The InstagramEventsJob uses a Redis mutex with a key based on sender_id
and ig_account_id to prevent race
conditions. However, all test factories were using the same hardcoded
sender_id: 'Sender-id-1', causing multiple
test instances to compete for the same mutex lock when running in
parallel.
### The Solution:
- Updated all Instagram event factories to generate unique sender IDs
using SecureRandom.hex(4)
- Modified test stubs and expectations to work with dynamic sender IDs
instead of hardcoded values
- Ensured each test instance gets its own unique mutex key, eliminating
lock contention
## Summary
- handle Twilio failures per contact when running one-off SMS campaigns
- rescue errors in WhatsApp and generic SMS one-off campaigns so they
continue
- add specs confirming campaigns continue sending when a single contact
fails
fixes: https://github.com/chatwoot/chatwoot/issues/9000
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
This update adds support to the coexistence method to Embedded Whatsapp,
allowing users to add their existing whatsapp business number in order
to use it in both places(chatwoot and whatsapp business) at the same
time.
This update require some changes in the permissions for the Meta App, as
described in the Meta Oficial Docs, I'll leave this listed below:
- **history** — describes past messages the business customer has
sent/received
- **smb_app_state_sync** — describes the business customer's current and
new contacts
- **smb_message_echoes** — describes any new messages the business
customer sends with the WhatsApp Business app after having been
onboarded
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
Co-authored-by: Sojan Jose <sojan@pepalo.com>
Co-authored-by: Tanmay Deep Sharma <32020192+tds-1@users.noreply.github.com>
# 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: Muhsin Keloth <muhsinkeramam@gmail.com>
Database CPU utilization was spiking due to expensive notification COUNT
queries. Analysis revealed two critical issues:
1. Missing database index: Notification count queries were performing
table scans without proper indexing
2. Duplicate WHERE clauses: SQL queries contained redundant read_at IS
NULL conditions, causing unnecessary query complexity
### Root Cause Analysis
The expensive queries were:
```
-- 41.61 calls/sec with duplicate condition
SELECT COUNT(*) FROM "notifications"
WHERE "notifications"."user_id" = $1
AND "notifications"."account_id" = $2
AND "notifications"."snoozed_until" IS NULL
AND "notifications"."read_at" IS NULL
AND "notifications"."read_at" IS NULL -- Duplicate!
```
This was caused by a logic error in NotificationFinder#unread_count
introduced in commit cd06b2b33 (PR #8907). The method assumed
@notifications contained all notifications, but @notifications was
already filtered to unread notifications in most cases.
### The Default Query Flow:
1. Frontend calls: NotificationsAPI.getUnreadCount() →
/notifications/unread_count
2. No parameters sent, so params = {}
3. NotificationFinder setup:
- find_all_notifications: WHERE user_id = ? AND account_id = ?
- filter_snoozed_notifications: WHERE snoozed_until IS NULL
- filter_read_notifications: WHERE read_at IS NULL (because
type_included?('read') is false)
4. unread_count called: Adds another WHERE read_at IS NULL
----
### Solution
1. Added Missing Database Index
- Index: (user_id, account_id, snoozed_until, read_at)
2. Fixed Duplicate WHERE Clause Logic
The term "sorcerer’s apprentice mode" is defined as a bug in a protocol
where, under some circumstances, the receipt of a message causes
multiple messages to be sent, each of which, when received, triggers the
same bug. - RFC3834
Reference: https://github.com/chatwoot/chatwoot/pull/9606
This PR:
- Adds an auto_reply attribute to message.
- Adds an auto_reply attribute to conversation.
- Disable conversation_created / conversation_opened event if auto_reply
is set.
- Disable message_created event if auto_reply is set.
---------
Co-authored-by: Muhsin Keloth <muhsinkeramam@gmail.com>
On Aug 2, we had a P0 because of a sudden spike in Sidekiq jobs. The
queue went up to 100k jobs and workers scaled from 400 threads to 1000+.
Most of the jobs were HookJobs, and a large chunk of them were for
Linear but they weren’t doing anything useful.
Turns out, whenever there’s an update on a contact or conversation, we
were triggering all account-level hooks without checking if the event
was relevant. So if someone did a bulk import or ran an update, it would
enqueue a huge number of unnecessary jobs.
This PR adds two checks before enqueuing:
- Whether the hook is active
- Whether the event is relevant for that hook
Add bounced emails to the conversation thread.
Fix Gmail bounce detection by checking the X-Failed-Recipients header.
Currently, bounced emails are rejected as auto-replies, which causes
support agents to miss important delivery failure context. This PR
ensures bounced messages are correctly added to the thread, preserving
visibility for the support team.
# Pull Request Template
## Description
Fixes
[CW-4620](https://linear.app/chatwoot/issue/CW-4620/rethinking-custom-domains-in-chatwoot)
<img width="642" height="187" alt="Screenshot 2025-07-29 at 8 17 44 PM"
src="https://github.com/user-attachments/assets/ad2f5dac-4b27-4dce-93ca-6cbba74443fb"
/>
## Type of change
- [x] New feature (non-breaking change which adds functionality)
## How Has This Been Tested?
## 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
- [x] 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: Vishnu Narayanan <iamwishnu@gmail.com>
Co-authored-by: Pranav <pranavrajs@gmail.com>
Co-authored-by: Pranav <pranav@chatwoot.com>
This PR migrates the legacy OpenAI integration (where users provide
their own API keys) from using hardcoded `https://api.openai.com`
endpoints to use the configurable `CAPTAIN_OPEN_AI_ENDPOINT` from the
captain configuration. This ensures consistency across all OpenAI
integrations in the platform.
## Changes
- Updated `lib/integrations/openai_base_service.rb` to use captain
endpoint config
- Updated `enterprise/app/models/enterprise/concerns/article.rb` to use
captain endpoint config
- Removed unused `enterprise/lib/chat_gpt.rb` class
- Added tests for endpoint configuration behavior
