What it does
If your support promise is "a reply within an hour", you need two numbers: the median, which says what a typical customer sees, and the share inside the hour, which says how often the promise holds. Intercom's reports show both, but only to the people who open them.
The Conversations API already returns each conversation's reply times. Each weekday the agent reads them over a rolling seven days and posts the median as a tile. Conversations closed without any teammate reply, often the ones Fin or another bot handled, are left out, so the number is about your team.
How it works
1Search the last 7 days
POST /conversations/searchforcreated_atgreater than now minus 7 days, withper_page: 150andstarting_afterfor paging. Drop anything your team started (an outbound message has asource.delivered_asother thancustomer_initiated).- 2
2Read the reply clock
Each conversation carries
statistics.time_to_admin_reply(seconds until the first teammate reply, with time outside office hours taken out) andstatistics.first_admin_reply_at. Skip closed conversations with no teammate reply; open ones with none are still waiting. - 3
3Compute the numbers
Median and 90th percentile of the reply time, and the share answered within your target (default 60 minutes). Split by team inbox with
team_assignee_idif you have more than one. - 4
4List who is waiting
Every open conversation with no teammate reply that is already past the target, longest wait first, with the inbox and a short description of what the customer asked.
- 5
5Post the median
Headline: median minutes to first reply, against yesterday's run. Glance: the share within target as a progress bar, with the 90th percentile and the waiting count.
Post it to the board
With
BRIEFKIN_TOKENset, the agent sends the result as one card that updates in place every run. Without it, you get the report in the chat.
Install and run it
Pick your agent, where the data comes from and how to schedule it; the commands below follow.
Install the skill
One command. It puts the SKILL.md in ~/.claude/skills and in ~/.agents/skills, which Codex, Cursor, Gemini CLI, Copilot CLI and OpenCode read too.
shell
curl -fsSL https://briefkin.dev/skills/intercom-first-response-time/install | shRun it
Ask Claude Code:
prompt
Run the intercom-first-response-time skillOr paste the prompt instead of installing:
prompt
Using the Intercom Conversations API, search for conversations created in the last 7 days that customers started. From each conversation's statistics take time_to_admin_reply, skipping closed conversations that never got a teammate reply. Give me the median and 90th percentile first response time and the share answered within 60 minutes, overall and by team inbox. Then list open conversations with no teammate reply that have already waited longer than 60 minutes, longest first, with a short description of what the customer asked.Put it on a schedule
Weekdays at 9:00, from cron on any machine where Claude Code is signed in:
crontab
0 9 * * 1-5 claude -p "Run the intercom-first-response-time skill"Add it with
crontab -e. A laptop asleep at that time runs it when it wakes if you use launchd or Task Scheduler instead; GitHub Actions works for the API route.Send it to your board
Connect Briefkin once with
curl -fsSL https://briefkin.dev/install | shand every run lands on your board as a stat tile. How setup works
View the SKILL.mdRaw file
SKILL.md---
name: intercom-first-response-time
description: "Each weekday: median time to first teammate reply in Intercom over the last 7 days, the share answered within your target, and who is still waiting. Use when the user asks for this, or when it runs on a schedule."
---
# Intercom first response time
Each weekday: median time to first teammate reply in Intercom over the last 7 days, the share answered within your target, and who is still waiting.
If your support promise is "a reply within an hour", you need two numbers: the median, which says what a typical customer sees, and the share inside the hour, which says how often the promise holds. Intercom's reports show both, but only to the people who open them.
The Conversations API already returns each conversation's reply times. Each weekday the agent reads them over a rolling seven days and posts the median as a tile. Conversations closed without any teammate reply, often the ones Fin or another bot handled, are left out, so the number is about your team.
## Two ways to get the data
Use the API or MCP route when you have an MCP server, a CLI or an API token for the app. Otherwise, if you can drive a browser the user is signed in to (Claude in Chrome, a Playwright or Chrome DevTools MCP server), use the browser route. Say which one you used.
## Route A: API or MCP server
### What you need
- An Intercom access token with read access to conversations, in `INTERCOM_ACCESS_TOKEN`, or an Intercom connector your agent already uses.
- Your response target (for example 60 minutes) and whether it applies in office hours only.
If something is missing, say what and stop; do not invent numbers.
### Steps
1. **Search the last 7 days** (Intercom): `POST /conversations/search` for `created_at` greater than now minus 7 days, with `per_page: 150` and `starting_after` for paging. Drop anything your team started (an outbound message has a `source.delivered_as` other than `customer_initiated`).
2. **Read the reply clock**: Each conversation carries `statistics.time_to_admin_reply` (seconds until the first teammate reply, with time outside office hours taken out) and `statistics.first_admin_reply_at`. Skip closed conversations with no teammate reply; open ones with none are still waiting.
3. **Compute the numbers**: Median and 90th percentile of the reply time, and the share answered within your target (default 60 minutes). Split by team inbox with `team_assignee_id` if you have more than one.
4. **List who is waiting**: Every open conversation with no teammate reply that is already past the target, longest wait first, with the inbox and a short description of what the customer asked.
5. **Post the median**: Headline: median minutes to first reply, against yesterday's run. Glance: the share within target as a progress bar, with the 90th percentile and the waiting count.
## Route B: the browser
### What you need
- Signed in to Intercom as a teammate with reporting permission (the Responsiveness template is in Reports) and access to the team inboxes.
If something is missing, say what and stop; do not invent numbers.
### Steps
Start at https://app.intercom.com/. The user is signed in already; never type passwords or codes, and ask if a sign-in page appears.
1. **Open the Responsiveness report**: Intercom URLs carry your workspace ID, so start at the landing page and pick the workspace. Open **Reports**, then the report built from the **Responsiveness** template (create one from the template if nobody has). Set the time period at the top to the **last 7 days**.
2. **Read first response time**: Read the **First response time** chart: it shows the share of conversations in seven time buckets, so the share answered within your target is the sum of the buckets up to **60 minutes**. Read the **Median response time** chart for the median, and hover over the info icon on each chart to confirm it uses office hours and counts time assigned to bot the way you want.
3. **Split by team**: Add a report-level filter by **Team** for each inbox you run and note the median and the within-target share per team. Clear the filter afterwards.
4. **Find who is still waiting**: Open **Inbox**, choose the **Unassigned** or **All** view and sort by **Waiting since**, oldest first. Note every open conversation a customer started that has no teammate reply yet and has waited more than **60 minutes**, with the inbox and what the customer asked.
5. **Write the numbers up**: Median first response, the share within target, the per-team split and the waiting list, longest wait first.
Note: The report shows buckets, not percentiles, so the 90th percentile the API route gives is not available here; say which bucket holds the slowest 10% instead. Office hours and bot time follow the chart settings, so check them before comparing with yesterday.
## The report
- **Median first response** over the last 7 days, and the change since yesterday.
- **Within target**: the share answered inside your target, and the 90th percentile.
- **By inbox**, when you have more than one team.
- **Still waiting**: open conversations past the target with no teammate reply.
## Post it to Briefkin
When `BRIEFKIN_TOKEN` is set, or `~/.config/briefkin/env` exists (the installer from https://briefkin.dev/docs writes it), also send the result to the user's Briefkin board as one card. Keep the `item_key` the same on every run so the card updates in place, use a new `event_id` per run, and put today's real numbers in place of the example ones. If the request fails, say so once and carry on.
```bash
[ -n "${BRIEFKIN_TOKEN:-}" ] || . ~/.config/briefkin/env
curl -sS -X POST "${BRIEFKIN_URL:-https://briefkin.dev}/v1/events" \
-H "authorization: Bearer $BRIEFKIN_TOKEN" \
-H "content-type: application/json" \
-d @- <<'JSON'
{
"v": 1,
"event_id": "intercom-first-response-time/<date of this run, e.g. 2026-10-16>",
"item_key": "intercom-first-response-time",
"project": "Acme",
"source": "Intercom first response time skill",
"kind": "metric",
"title": "Median first response 38 min, 81% within an hour",
"metric": {
"label": "Median first response",
"value": 38,
"unit": "min",
"better": "lower",
"period_start": "<start of the period, ISO 8601>",
"period_end": "<end of the period, ISO 8601>"
},
"glance": [
{
"type": "Progress",
"props": {
"label": "Answered within 1 hour",
"value": 81,
"text": "583 of 720 conversations",
"tone": "ok"
}
},
{
"type": "Grid",
"props": {
"columns": 2
},
"children": [
{
"type": "Stat",
"props": {
"label": "90th percentile",
"value": 3.2,
"unit": "h"
}
},
{
"type": "Stat",
"props": {
"label": "Waiting past target",
"value": 6,
"tone": "warn"
}
}
]
}
],
"widget": [
{
"type": "Table",
"props": {
"columns": [
"Inbox",
"Conversations",
"Median",
"Within 1 h"
],
"align": [
"left",
"right",
"right",
"right"
],
"rows": [
[
"Billing",
214,
"24 min",
"89%"
],
[
"Technical",
301,
"51 min",
"71%"
],
[
"Sales questions",
205,
"33 min",
"86%"
]
]
}
},
{
"type": "Table",
"props": {
"columns": [
"Waiting for a first reply",
"Inbox",
"Waiting"
],
"align": [
"left",
"left",
"right"
],
"rows": [
[
"API key stopped working after rotation",
"Technical",
"4 h 10 min"
],
[
"Needs a W-9 for vendor setup",
"Billing",
"2 h 45 min"
],
[
"Webhook signatures fail to verify",
"Technical",
"2 h 20 min"
],
[
"Asks about moving to annual billing",
"Sales questions",
"1 h 50 min"
],
[
"Bulk import drops the last row",
"Technical",
"1 h 25 min"
],
[
"Seat count wrong after removing users",
"Billing",
"1 h 5 min"
]
]
}
},
{
"type": "Callout",
"props": {
"text": "Technical is the slow inbox: a 51 minute median and 3 of the 6 conversations waiting past target.",
"tone": "warn"
}
}
],
"size": "md"
}
JSON
```
Widgets come from the Briefkin catalog (https://briefkin.dev/v1/catalog). Without a token, just give the user the report.
Set `period_start` and `period_end` to the window the number covers: the tile names it from them ("Last 7 days", "Today", "This month"). For a count right now (open items, a balance), send `"period_label": ""` so no period shows; for another name, send it as `period_label` (at most 24 characters).
## Run it on a schedule
Suggested: Weekdays at 9:00 (`0 9 * * 1-5`). From cron, with the agent signed in on that machine:
```bash
0 9 * * 1-5 claude -p "Run the intercom-first-response-time skill"
# Codex: codex exec, Gemini CLI: gemini -p, Copilot CLI: copilot -p, OpenCode: opencode run
```
Or use the agent's own scheduler (Claude Code routines, Codex automations, Claude or ChatGPT scheduled tasks): https://briefkin.dev/skills/intercom-first-response-time#schedule
Source: https://briefkin.dev/skills/intercom-first-response-time
What your agent needs
An Intercom access token with read access to conversations, in
INTERCOM_ACCESS_TOKEN, or an Intercom connector your agent already uses.Your response target (for example 60 minutes) and whether it applies in office hours only.
Connecting Intercom
- Intercom MCP server, Intercom's official remote server at
https://mcp.intercom.com/mcp(https://mcp.eu.intercom.com/mcpfor EU workspaces; Australian workspaces aren't supported yet). Intercom is in Claude's connector directory for US workspaces, and on Team and Enterprise plans an Owner adds it. Its tools search and fetch conversations and contacts (docs). - REST API with an access token. Create an app in the Intercom Developer Hub, store its access token in
INTERCOM_ACCESS_TOKEN, and callhttps://api.intercom.io(api.eu.intercom.ioorapi.au.intercom.ioby region) with anIntercom-Versionheader. Give the app Read conversations and Read and list users and companies.
The MCP server includes a few write tools (internal notes, Help Center articles), so for unattended reports an app token with read permissions only is the tighter option.
Questions
Does it count nights and weekends?
Not by default. time_to_admin_reply leaves out time outside the office hours set in Intercom, the same way Intercom's own reports do. If your promise is wall-clock, the skill can use first_admin_reply_at minus created_at instead.
Can it warn me while the queue is slow, not just once a day?
The 7-day median moves slowly on purpose. For a live alarm, run only the waiting check every 15 minutes during office hours and post an action card when a conversation passes the target.
Do I need Briefkin to use the Intercom first response time skill?
No. The skill runs with your own agent and gives you the report in the chat. With a Briefkin token set, each run also posts it to your board as a stat tile, so you can glance at it instead of asking again.
Which agents can run it?
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI, OpenCode, Windsurf and Amp all read SKILL.md files. The install command puts it in ~/.claude/skills and ~/.agents/skills, where they look. ChatGPT, the Claude app and Grok take the prompt instead. Pick your agent under "Install and run it" and the commands change to match.
Can it work without an API or MCP server, from my browser?
Yes. Switch the data source to "Your browser" and the steps change to the Intercom pages themselves: the agent opens app.intercom.com in a browser you are signed in to and reads the numbers off the page. Claude Code, Codex, Cursor, Claude app and ChatGPT can drive a browser out of the box; Gemini CLI, Copilot CLI and OpenCode need a Playwright or Chrome DevTools MCP server first. It is slower than the API and needs you signed in, so it suits a report you run by hand or from a scheduler on your own computer.