feat(dashboard, api-service): Filter conversations by agent; add recent UI & CTA#10767
Merged
feat(dashboard, api-service): Filter conversations by agent; add recent UI & CTA#10767
Conversation
Expose agentId filtering end-to-end and introduce a compact recent conversations UI with upgrade CTA. - API: add agentId to ConversationFilters and include it in getConversationsList query params. - DAL: apply agentId (_agentId) filter in ConversationRepository queries. - UI: refactor RecentConversationsSection to fetch recent conversations for an agent, render skeleton/error/empty states, conversation items, and format timestamps; gate detailed list behind IS_ENTERPRISE and show ConversationsUpgradeCta for non-enterprise users. - Conversations page: gate main conversations content behind IS_ENTERPRISE and split into EnterpriseConversationsContent. - Add ConversationsUpgradeCta component (compact/default variants) to prompt upgrade/contact-sales and link docs. - Update submodule reference (.source). These changes enable agent-scoped conversation listing and a clear upgrade path for non-enterprise installs.
✅ Deploy Preview for dashboard-v2-novu-staging canceled.
|
Contributor
|
Hey there and thank you for opening this pull request! 👋 We require pull request titles to follow specific formatting rules and it looks like your proposed title needs to be adjusted. Your PR title is: Requirements:
Expected format: Details: PR title must end with 'fixes TICKET-ID' (e.g., 'fixes NOV-123') or include ticket ID in branch name |
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
✅ Files skipped from review due to trivial changes (1)
📝 WalkthroughWalkthroughAdds optional agent-scoped conversation filtering across API and DAL, updates MongoDB indexes, gates conversations UI behind an enterprise flag with an upgrade CTA for non-enterprise, refactors recent-conversations to fetch/render agent-specific items, and updates the Changes
Sequence DiagramsequenceDiagram
participant UI as Dashboard (UI)
participant API as API Layer\n(`apps/dashboard/src/api/conversations.ts`)
participant DAL as Repository\n(`libs/dal/.../conversation.repository.ts`)
participant DB as MongoDB
UI->>UI: Check IS_ENTERPRISE
alt Enterprise
UI->>API: GET /conversations?limit=5&agentId=...
API->>DAL: listConversations({ agentId, limit:5, ... })
DAL->>DB: Query {_environmentId, _agentId, ...}
DB-->>DAL: Documents
DAL-->>API: Conversation list
API-->>UI: Filtered results
UI->>UI: Render recent list with links/status/subscriber
else Non-Enterprise
UI->>UI: Render ConversationsUpgradeCta
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Suggested labels
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx`:
- Around line 161-184: RecentConversationsSkeleton currently renders 4
placeholder rows causing a layout shift; change it to render
RECENT_CONVERSATIONS_DISPLAY_LIMIT placeholders instead: import or reference the
RECENT_CONVERSATIONS_DISPLAY_LIMIT constant and use it as the length in
Array.from inside RecentConversationsSkeleton (keep the existing key pattern
like `skeleton-${index}`), so the skeleton count matches the real display limit
of 5.
- Around line 104-118: RecentConversationItem currently builds a detailPath and
renders a Link even when environmentSlug is missing, resulting in a dead "#"
link; update RecentConversationItem to avoid a non-functional interactive
element by either early-returning (do not render the item) when environmentSlug
is undefined or render a non-interactive container (e.g., a div/buttonless
element with the same classes) instead of Link when environmentSlug is falsy;
change usage of detailPath and the Link import accordingly so the clickable Link
is only used when environmentSlug is present and the non-interactive fallback
preserves visual styles but removes hover/focus interactivity.
In `@apps/dashboard/src/components/conversations/conversations-upgrade-cta.tsx`:
- Around line 86-90: The current code nests a LinkButton inside react-router's
<Link> for an absolute URL, creating nested interactive elements; change this to
either render a plain anchor for the external URL or make LinkButton render the
anchor directly via its asChild prop: replace the outer <Link
to="https://docs.novu.co/agents/overview" ...> wrapper with a standard <a
href="https://docs.novu.co/agents/overview" target="_blank" rel="noreferrer
noopener"> or keep LinkButton but pass asChild so LinkButton renders the <a>
itself (keeping the RiBookMarkedLine icon and size="sm" props) to avoid invalid
nested <a>/<button> elements and fix accessibility.
In `@libs/dal/src/repositories/conversation/conversation.repository.ts`:
- Around line 249-251: The Conversation repository adds an _agentId filter in
the query (see the if (agentId) { query._agentId = agentId; } block) but the
Conversation schema lacks a compound index that covers queries scoped by
_environmentId plus _agentId; add a compound index on the Conversation model
like { _environmentId: 1, _agentId: 1, _id: 1 } to avoid collection scans for
listConversations and, if listConversations is commonly sorted by other fields
such as createdAt or lastActivityAt, also add indexes like { _environmentId: 1,
_agentId: 1, createdAt: 1 } or { _environmentId: 1, _agentId: 1, lastActivityAt:
1 } in the Conversation schema definition so queries using the query._agentId
predicate and various sortBy values are covered.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: c25f5e2f-cb74-4fd8-8f9b-4cb24a43c072
📒 Files selected for processing (6)
.sourceapps/dashboard/src/api/conversations.tsapps/dashboard/src/components/agents/recent-conversations-section.tsxapps/dashboard/src/components/conversations/conversations-content.tsxapps/dashboard/src/components/conversations/conversations-upgrade-cta.tsxlibs/dal/src/repositories/conversation/conversation.repository.ts
Refactor RecentConversationItem to extract base and interactive classNames, render a non-interactive div when environmentSlug is missing, and only wrap with a Link when environmentSlug exists. Use RECENT_CONVERSATIONS_DISPLAY_LIMIT for skeleton item count. Update ConversationsUpgradeCta to use LinkButton asChild with a native anchor for the external docs link and remove the unused Link import. Add agent-scoped indexes to the conversation schema (_environmentId + _agentId with _id, createdAt, lastActivityAt) to improve query performance for agent-scoped operations.
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
apps/dashboard/src/components/agents/recent-conversations-section.tsx (2)
22-26: Minor: duplicateduseEnvironment()call.
currentEnvironmentis read here and again inRecentConversationsContent(line 57). Since the parent already computesconversationsPath, consider passingcurrentEnvironment?.slug(orconversationsPath) down as a prop to avoid re-subscribing to the context in the child.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx` around lines 22 - 26, The component calls useEnvironment() twice (once here to build conversationsPath and again inside RecentConversationsContent), causing a duplicate context subscription; pass the computed value down instead: compute currentEnvironment?.slug (or conversationsPath) in this component (where useEnvironment() is already used), then add a prop to RecentConversationsContent (e.g., environmentSlug or conversationsPath) and update RecentConversationsContent to consume that prop instead of calling useEnvironment() itself (update usages around buildRoute/ROUTES.ACTIVITY_CONVERSATIONS accordingly).
207-211:conversation._agentIdfallback will render as a raw ObjectId.If neither
agent.agent?.namenoragent?.idis present, the UI displays the raw ObjectId from_agentIdwhich is not user-friendly. Since the parentRecentConversationsSectionalready has the fullAgentResponsein scope, consider threading the agent's display name down toRecentConversationItemand using it as the fallback (all listed conversations belong to that agent anyway).🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx` around lines 207 - 211, getAgentName currently falls back to rendering conversation._agentId (a raw ObjectId); instead, pass the parent AgentResponse's display name down from RecentConversationsSection into each RecentConversationItem and use that as the final fallback. Update the caller in RecentConversationsSection to provide a prop (e.g., agentDisplayName or agent.name) to RecentConversationItem, then change getAgentName(conversation: ConversationDto) to accept an optional agentDisplayName parameter and return agent?.agent?.name ?? agent?.id ?? agentDisplayName ?? 'agent'; ensure function and prop names (getAgentName, RecentConversationItem, RecentConversationsSection, AgentResponse, conversation._agentId) are used consistently.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx`:
- Around line 59-62: The useFetchConversations call is passing the wrong agent
identifier; update the filters argument in the useFetchConversations invocation
(the call that uses RECENT_CONVERSATIONS_DISPLAY_LIMIT) to use agent._id instead
of agent.identifier so the agentId filter matches the ObjectId stored in the
conversation schema.
---
Nitpick comments:
In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx`:
- Around line 22-26: The component calls useEnvironment() twice (once here to
build conversationsPath and again inside RecentConversationsContent), causing a
duplicate context subscription; pass the computed value down instead: compute
currentEnvironment?.slug (or conversationsPath) in this component (where
useEnvironment() is already used), then add a prop to RecentConversationsContent
(e.g., environmentSlug or conversationsPath) and update
RecentConversationsContent to consume that prop instead of calling
useEnvironment() itself (update usages around
buildRoute/ROUTES.ACTIVITY_CONVERSATIONS accordingly).
- Around line 207-211: getAgentName currently falls back to rendering
conversation._agentId (a raw ObjectId); instead, pass the parent AgentResponse's
display name down from RecentConversationsSection into each
RecentConversationItem and use that as the final fallback. Update the caller in
RecentConversationsSection to provide a prop (e.g., agentDisplayName or
agent.name) to RecentConversationItem, then change getAgentName(conversation:
ConversationDto) to accept an optional agentDisplayName parameter and return
agent?.agent?.name ?? agent?.id ?? agentDisplayName ?? 'agent'; ensure function
and prop names (getAgentName, RecentConversationItem,
RecentConversationsSection, AgentResponse, conversation._agentId) are used
consistently.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 218e3b28-c631-4893-ba5d-773522edf890
📒 Files selected for processing (3)
apps/dashboard/src/components/agents/recent-conversations-section.tsxapps/dashboard/src/components/conversations/conversations-upgrade-cta.tsxlibs/dal/src/repositories/conversation/conversation.schema.ts
Comment on lines
+59
to
+62
| const { conversations, isLoading, isError } = useFetchConversations({ | ||
| limit: RECENT_CONVERSATIONS_DISPLAY_LIMIT, | ||
| filters: { agentId: agent.identifier }, | ||
| }); |
Contributor
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Inspect AgentResponse shape and how _agentId is populated on conversations.
fd -t f 'agents.ts' apps/dashboard/src/api | xargs -I{} sh -c 'echo "=== {} ==="; cat {}'
echo "---- agent entity / schema ----"
fd -t f 'agent.entity.ts|agent.schema.ts' libs/dal/src | xargs -I{} sh -c 'echo "=== {} ==="; cat {}'
echo "---- conversation creation / _agentId assignment ----"
rg -nP --type=ts -C3 '_agentId\s*[:=]'Repository: novuhq/novu
Length of output: 30067
🏁 Script executed:
fd -t f 'conversations.ts' apps/dashboard/src -type fRepository: novuhq/novu
Length of output: 225
🏁 Script executed:
rg -n 'useFetchConversations' apps/dashboard/src --type=ts --type=tsx -A 5Repository: novuhq/novu
Length of output: 82
🏁 Script executed:
rg -n 'const.*fetchConversations\|export.*useFetchConversations' apps/dashboard/src --type=ts --type=tsx -B 2 -A 10Repository: novuhq/novu
Length of output: 82
🏁 Script executed:
fd -type f 'conversations' apps/dashboard/srcRepository: novuhq/novu
Length of output: 225
🏁 Script executed:
rg 'useFetchConversations' apps/dashboard/src -g '*.ts' -g '*.tsx' -A 5Repository: novuhq/novu
Length of output: 3160
🏁 Script executed:
rg 'fetchConversations|useFetchConversations' apps/dashboard/src -g '*.ts' -g '*.tsx' -B 2 -A 10Repository: novuhq/novu
Length of output: 9446
🏁 Script executed:
cat apps/dashboard/src/api/conversations.tsRepository: novuhq/novu
Length of output: 4696
Use agent._id instead of agent.identifier for the agentId filter.
The conversation schema stores agent references as _agentId: Schema.Types.ObjectId, but you're filtering with agent.identifier (a string slug). The API will send ?agentId=<identifier-slug> to the backend, which tries to match it against an ObjectId field. This causes Mongoose to either throw a cast error or silently return zero results, leaving the Recent Conversations list empty.
Change line 61 to:
filters: { agentId: agent._id },🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/dashboard/src/components/agents/recent-conversations-section.tsx` around
lines 59 - 62, The useFetchConversations call is passing the wrong agent
identifier; update the filters argument in the useFetchConversations invocation
(the call that uses RECENT_CONVERSATIONS_DISPLAY_LIMIT) to use agent._id instead
of agent.identifier so the agentId filter matches the ObjectId stored in the
conversation schema.
Contributor
Author
There was a problem hiding this comment.
But the backend converts the identifier to _id
Contributor
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Look for backend conversion of agentId/identifier to _id in conversation-related code
rg -n 'identifier|agentId' apps/api/src --type=ts -g '*conversation*' -A 5 -B 2
echo "---- Check use-cases or services that handle agentId ----"
rg -rn 'agentId' apps/api/src --type=ts -A 5 -B 2
Length of output: 95710
🏁 Script executed:
#!/bin/bash
# Find the list-conversations usecase to see how agentId is processed
rg -rn 'agentId' apps/api/src --type=ts -g '*conversation*' -A 5 -B 2
echo "---- list conversations usecase ----"
fd -t f 'list-conversations' apps/api/src | xargs -I{} sh -c 'echo "=== {} ==="; cat {}'
echo "---- conversations controller ----"
fd -t f 'conversations.controller.ts' apps/api/src | xargs -I{} sh -c 'echo "=== {} ==="; cat {}'
Length of output: 1897
🏁 Script executed:
#!/bin/bash
# Broader search for conversations listing and agentId handling
rg -rn 'agentId' apps/api/src --type=ts -A 5 -B 2
echo "==== conversations controller candidates ===="
fd -t f 'conversation' apps/api/src --extension ts | grep -v 'agent-conversation'
echo "==== list conversations ===="
rg -rn 'listConversations\|list_conversations\|getConversationsList\|ConversationFilters' apps/api/src --type=ts -A 5 -B 2
Length of output: 92861
🏁 Script executed:
#!/bin/bash
# Find the conversations listing API endpoint in apps/api
fd -t f --extension ts 'conversation' apps/api/src | grep -v 'agent-conversation\|e2e\|spec'
echo "==== search for agentId in non-agents context ===="
rg -rn 'agentId' apps/api/src --type=ts --iglob '!*agent*' -A 5 -B 3
echo "==== dal listConversations call sites ===="
rg -rn 'listConversations' --type=ts -A 5 -B 3
Length of output: 789
🏁 Script executed:
#!/bin/bash
cat libs/dal/src/repositories/conversation/conversation.repository.ts
echo "==== Find conversations API controller/usecase ===="
fd -t f --extension ts . apps/api/src | xargs grep -l 'agentId\|agent_id\|_agentId' 2>/dev/null | grep -v 'agent-conversation\|bridge-executor\|chat-sdk\|agent-inbound\|agent-config'
echo "==== conversations usecase files ===="
fd -t f --extension ts . apps/api/src/app | xargs grep -l 'ConversationFilters\|conversations.*usecase\|usecase.*conversation' 2>/dev/null
Length of output: 8864
🏁 Script executed:
#!/bin/bash
# Find conversations API endpoint/controller outside agents module
rg -rn 'getConversationsList\|conversations.*list\|list.*conversations' apps/api/src --type=ts -B 2 -A 5
echo "==== search for conversation routes ===="
rg -rn "'/conversations'\|\"conversations\"" apps/api/src --type=ts -B 2 -A 3
echo "==== Check what URL dashboard calls for conversations ===="
cat apps/dashboard/src/api/conversations.ts
Length of output: 4741
🏁 Script executed:
#!/bin/bash
# Find the /conversations endpoint handler
rg -rn "conversations" apps/api/src --type=ts -g '*controller*' -B 2 -A 5
echo "==== search for conversation module ===="
fd -t d 'conversation' apps/api/src
echo "==== check all controllers for conversations route ===="
fd -t f 'controller' apps/api/src --extension ts | xargs grep -l 'conversation' 2>/dev/null
Length of output: 194
🏁 Script executed:
#!/bin/bash
# Search across all apps for the conversations GET endpoint
rg -rn "Get.*conversations\|conversations.*Get\|@Get.*conver\|route.*conversation\|conversation.*route" --type=ts -B 2 -A 5
echo "==== find conversations endpoint in all apps ===="
rg -rn "'/conversations'\|\"\/conversations\"\|'/v1/conversations'\|v1.*conversations" --type=ts -B 2 -A 5
Length of output: 210
🏁 Script executed:
#!/bin/bash
# Search for conversations endpoint across all apps
rg -rn "conversations" --type=ts -g '*controller*' -B 2 -A 5
echo "==== find route /conversations ===="
rg -rn "conversations" --type=ts -g '*router*' -B 2 -A 5
echo "==== search all TS files for conversations endpoint ===="
rg -rn "@Get\('conversations'\)" --type=ts -B 2 -A 5
echo "==== look for usecase that handles list conversations with agentId conversion ===="
rg -rn "agentId" --type=ts -g '!**/agents/**' -B 3 -A 5
Length of output: 75597
@scopsy Thanks for the clarification — you’re right that several agent endpoints take an identifier and resolve it to _id server‑side. For conversations, though, the listing path expects the agent’s ObjectId under query param n, and the DAL filters on _n (ObjectId).
Concretely:
- Conversation schema uses _n: ObjectId
- ConversationRepository.listConversations takes n and sets query._n = n
- Dashboard API client appends n when provided in filters and ConversationDto exposes _n
Given that, this component should pass the agent’s _id under n rather than the textual identifier under agentId. Suggested change in apps/dashboard/src/components/agents/recent-conversations-section.tsx:
const { conversations, isLoading, isError } = useFetchConversations({
limit: RECENT_CONVERSATIONS_DISPLAY_LIMIT,
- filters: { agentId: agent.identifier },
+ filters: { n: agent._id },
});This aligns the UI with the conversations API contract and will reliably scope results to the agent. If you prefer keeping a friendlier prop name (agentId) in the hook, we can also translate agentId → n inside useFetchConversations, but today the API client already speaks in terms of n, so wiring it directly here is the smallest change.
🧠 Learnings used
Learnt from: CR
Repo: novuhq/novu PR: 0
File: .cursor/rules/pullrequest.mdc:0-0
Timestamp: 2026-03-24T13:18:35.551Z
Learning: Pull request title format: `type(scope): Description fixes NOV-<ticket-id>` (omit `fixes NOV-XXX` when no Linear ticket is in context). Example: `feat(dashboard): add workflow trigger button fixes NOV-123`
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Expose agentId filtering end-to-end and introduce a compact recent conversations UI with upgrade CTA.
These changes enable agent-scoped conversation listing and a clear upgrade path for non-enterprise installs.
What changed? Why was the change needed?
What changed
Agent-scoped conversation listing was implemented end-to-end: API and DAL accept an optional agentId filter, the dashboard UI fetches and displays recent conversations scoped to an agent, and a compact Recent Conversations panel plus an upgrade CTA were added to handle non-enterprise/self-hosted installs. The change enables agent-specific conversation views and provides a clear upgrade/contact-sales path where full conversation listing is enterprise-only.
Affected areas
Key technical decisions
Testing
No automated tests were added; manual verification is recommended for agent-scoped listing correctness, UI loading/error/empty states, enterprise gating behavior, CTA navigation for self-hosted vs SaaS, and telemetry firing on CTA clicks.
Screenshots
Expand for optional sections
Related enterprise PR
Special notes for your reviewer