HappyFox Chat
Helpdesk & ITSMRead live chat data from HappyFox Chat, covering transcripts, offline messages, visitors with full browsing attribution, agents, and canned responses. Agents pull the conversation history a follow-up actually needs.
What This Integration Enables
Every documented endpoint on this API is a GET. There are no writes, no replies, no status changes, nothing that reaches back into a live conversation. That is unusual enough to be the whole frame for this page. HappyFox Chat is not a place your agents act. It is a place they read, and everything they do with what they read happens in another system. Once you accept that, the design question stops being "what can the agent do here" and becomes "what should be allowed to leave", which is a better question anyway.
What can be read is genuinely rich. Transcripts carry the ordered message list with typed events, so an agent can tell an answered question from a visitor who went quiet, plus the profile, department, agents, first message, visitor rating, waiting time and average response time. Offline messages carry the leads that arrived when nobody was on. Visitors carry the deepest record in the API: IP address, timezone, UTM parameters, browser, operating system, device family, and the page history with titles, URLs, referrers and viewport dimensions. Agents, departments, profiles, canned responses, custom field definitions and proactive chat rules fill in the rest, so a report can resolve an ID into a name and a flow can reuse the team's approved phrasing instead of inventing its own. HappyFox Chat's webhooks are configured in the console rather than through the API, so both triggers poll, baseline silently on the first cycle, and deduplicate by record ID.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Finished chats that become tickets with the story intact
A chat ends. On New Chat Transcript fires with the summary, and the agent calls Get Chat Transcript for the full message list. It reads how the conversation closed, because a VisitorBecameInactive event at the end of an unanswered question means something very different from a resolved thread. Where the issue is open, the agent drafts a summary, pulls the matching approved wording with Get Canned Response, and files a ticket in Zendesk with the transcript attached. The agent who takes it reads what actually happened rather than a one-line subject.
Out-of-hours messages that reach a person before the queue does
A visitor leaves an offline message at 23:40 because nobody was available. On New Offline Message fires within the polling interval. The agent pulls the record with Get Offline Message, enriches it with the visitor record so the sales rep can see which pricing page they were on and where they came from, and creates the follow-up task in HubSpot. By the time the team is back, the lead is assigned and carries its own context.
Per-department chat quality reporting without the export ritual
A weekly flow pages through List Chat Transcripts, resolves each department and agent ID into a name with Get Department and Get Agent, and builds a breakdown of volume, waiting time, average response time and visitor rating per team. Because the API documents no filter, date range or sort parameters, the flow does the windowing itself and keeps its own rolling store. The support lead reads a trend rather than a screenshot, and the numbers come from the transcripts rather than from memory.
Human-in-Loop Highlight
A read-only connector concentrates all of its risk in the export, and this one reads the most confessional surface a company operates. People type things into a chat widget that they would never put in an email, because the box feels ephemeral and the person on the other end feels like one person. Account numbers, health details, a complaint naming a colleague, a screenshot of something they should not have pasted. The moment an agent summarises that transcript into a ticket, it moves the content into a system with a wider audience and a longer retention, and the visitor never agreed to that trip.
So the FlowRunner agent classifies before it copies. Most transcripts contain nothing sensitive and are filed automatically with the transcript attached, which is the whole point of running this unattended. When the classifier finds personal or sensitive content, the flow treats it as an automation exception and stops, sending the support lead a message that names the category without reproducing the content: "Transcript 8f21 from the billing widget contains what looks like a full card number typed by the visitor. I can file a redacted summary with the card digits removed, file nothing and leave it to you, or file it in full. Which?" The lead answers in Slack, the agent does exactly that and nothing else, and the decision is on the record. The agent reads everything. A person decides what travels.
Agent Capabilities
18 actionsTranscripts
2- List Chat Transcripts Lists completed chat transcripts newest first, each summarising the chat with its profile, department, agents, first message, visitor rating and timing statistics such as waiting time and average response time. Paginated with page and limit, capped at 30 records per page and 10 when no limit is given. The API documents no filter, date range or sort parameters, so a flow that needs a window keeps its own.
- Get Chat Transcript Retrieves a single transcript by ID with the full ordered message list. Entries carry a type such as VisitorSessionBegan, VisitorMessage, AgentJoined, AgentMessage or VisitorBecameInactive, the last of which reports how the conversation ended. Used to feed a complete conversation to an AI step for summarisation or quality scoring.
Offline Messages
2- List Offline Messages Lists the messages visitors left through the widget while no agent was available. These are the leads that would otherwise be found the next morning, so this is the usual source for an out-of-hours follow-up flow. Paginated with page and limit, capped at 30 records per page.
- Get Offline Message Retrieves a single offline message by ID with the visitor's contact details and the text they left. Used after a trigger or a list step to pull the full record before creating a ticket or a CRM task.
Agents
2- List Agents Lists the agents in the HappyFox Chat account. Agent IDs appear on transcripts, so this is how a reporting flow resolves the people named on a chat into full records. Paginated with page and limit, capped at 30 records per page.
- Get Agent Retrieves a single agent by ID. Used to attach a name and email to a transcript or a report row without paging the whole agent list.
Visitors
2- List Visitors Lists the visitors tracked by the chat widget with their contact details, repeat-visitor flag and last event timestamp. Paginated with page and limit, capped at 30 records per page.
- Get Visitor Retrieves a single visitor by ID, including the stats block with IP address, timezone, UTM parameters, browser, operating system, device family, and the page history they browsed with titles, URLs, referrers and viewport dimensions. This is the richest attribution source in the API, and the one worth deciding deliberately where to send.
Organisation
4- List Profiles Lists the chat profiles configured in the account. A profile is a widget installation, so its ID is what distinguishes chats arriving from different websites or products on a shared report. Paginated with page and limit, capped at 30 records per page.
- Get Profile Retrieves a single chat profile by ID. Used to resolve the profile referenced on a transcript when a flow routes by website or product.
- List Departments Lists the departments chats can be routed to. Department names appear on every transcript, so this is how a reporting flow builds a per-team breakdown. Paginated with page and limit, capped at 30 records per page.
- Get Department Retrieves a single department by ID. Used to confirm a department exists, or to attach its name to a report row carrying only the department ID.
Configuration
6- List Canned Responses Lists the canned responses agents can insert into a chat. Pulling them into a flow lets an AI step reuse the team's approved wording rather than inventing new phrasing. Paginated with page and limit, capped at 30 records per page.
- Get Canned Response Retrieves a single canned response by ID with its title and message body. Used to inject the exact approved wording into a reply sent through another channel.
- List Custom Fields Lists the custom fields defined on the widget's pre-chat and contact forms. Used to discover which extra data the widget collects before a flow tries to read those values off a visitor or transcript. Paginated with page and limit, capped at 30 records per page.
- Get Custom Field Retrieves a single custom field definition by ID. Used to confirm a field's identity before mapping its captured value into a CRM or ticketing system.
- List Triggered Chats Lists the proactive chat rules configured in the account. These are the rules that start a chat automatically when a visitor meets a condition, and transcripts record which chats they produced through the isTriggeredChat flag. Paginated with page and limit, capped at 30 records per page.
- Get Triggered Chat Retrieves a single proactive chat rule by ID. Used to audit which rule produced a triggered conversation when measuring the effect of proactive chat on conversion.
Triggers
2 triggersEvent Triggers
2- On New Chat Transcript Fires when a chat ends and its transcript is published, emitting the summary with its profile, departments, agents, first message, visitor rating and timing statistics. Follow it with Get Chat Transcript to read the full message list. Used to file a finished conversation as a ticket or a quality-review record. The first cycle records the transcripts that already exist without firing, so history is never replayed. Afterwards each new transcript fires once, oldest first, deduplicated by record ID. The trigger polls, because HappyFox Chat webhooks are configured in the console with no create or delete endpoint, and the interval can be set as low as 30 seconds.
- On New Offline Message Fires when a visitor leaves an offline message because no agent was available, emitting the record with the contact details and the text submitted. Used to raise a ticket or notify a sales channel the moment an out-of-hours lead arrives. The first cycle records the existing messages without firing, so history is never replayed. Afterwards each new message fires once, oldest first, deduplicated by record ID, and the polling interval can be set as low as 30 seconds.
Frequently Asked Questions
What can FlowRunner do with HappyFox Chat?
FlowRunner agents can run List Chat Transcripts, Get Chat Transcript, and List Offline Messages in HappyFox Chat, plus 15 more actions.
Does connecting HappyFox Chat to FlowRunner require OAuth?
No. HappyFox Chat connects to FlowRunner with an API key, no OAuth flow required.
Can HappyFox Chat trigger a FlowRunner workflow automatically?
Yes. HappyFox Chat supports 2 triggers that can start a FlowRunner workflow automatically.
Start building with HappyFox Chat
$100 in credits. No card required. Connect in minutes.