--- name: sentry-crash-free-sessions description: "Every 4 hours: the crash-free session rate of your newest release next to the previous one, with adoption, so an early number is read with care. Use when the user asks for this, or when it runs on a schedule." --- # Sentry crash-free sessions for the latest release Every 4 hours: the crash-free session rate of your newest release next to the previous one, with adoption, so an early number is read with care. Right after a release, the crash-free rate swings with every few hundred sessions, and it is easy to celebrate or panic too early. App stores, product leads and the on-call engineer all watch the same number, usually by refreshing the Releases page. The agent finds the newest release, reads its crash-free session rate from Sentry's release health data, puts it next to the previous release, and says what share of traffic is on the new version yet. A 99.1% on 3% of sessions is labelled early, not alarming, and the crashes behind the number are listed by users affected. ## 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 - Release health on: the SDK sets `release` and sends sessions (the mobile SDKs and recent JavaScript SDKs do by default). - A Sentry token with `org:read`, `project:read` and `event:read` in `SENTRY_AUTH_TOKEN`, or the Sentry MCP server. If something is missing, say what and stop; do not invent numbers. ### Steps 1. **Find the latest release** (Sentry): `GET /api/0/organizations/{org}/releases/?project=&per_page=2` returns the newest releases first. Keep the version and `dateCreated` of the latest and of the one before it. 2. **Read crash-free rates** (Sentry): `GET /api/0/organizations/{org}/sessions/?project=&field=crash_free_rate(session)&field=crash_free_rate(user)&field=sum(session)&query=release:&statsPeriod=24h&interval=4h`, once for each release, over the same window. 3. **Work out adoption** (Sentry): Divide the new release's `sum(session)` by all sessions in the project over the same window (the same call without the release filter). Under 10% adoption or 1,000 sessions, call the rate early. 4. **Find the crashes behind it** (Sentry): List the release's top crash issues with `GET .../issues/?query=release: error.unhandled:true&sort=user&limit=5`. 5. **Judge**: Healthy when the rate is within 0.3 points of the previous release, regressed when it is lower by more than that with at least 1,000 sessions, early when there is too little traffic to tell. ## Route B: the browser ### What you need - Signed in to Sentry as a member of the organization that owns the project; the Releases page shows health only for projects whose SDK sends sessions. If something is missing, say what and stop; do not invent numbers. ### Steps Start at https://sentry.io/releases/. The user is signed in already; never type passwords or codes, and ask if a sign-in page appears. 1. **Open Releases for the project**: The URL opens the Releases page of your last organization; switch organizations in the top left if needed. Pick the project (acme-ios) in the project selector, set the time range to **24H**, and set **Display** to `Sessions`. 2. **Read the two newest releases**: With the list sorted by date created, the top two rows are the latest release and the one before it. For each, note the version, **Crash Free Rate**, **Adoption** and the session count in the row. 3. **Open the latest release**: Click its version. On the release page read **Crash Free Session Rate**, **Crash Free User Rate** and **Session Count** from the table under the chart, and the **Sessions Adopted** graph for the share of traffic on it. 4. **List the crashes behind it**: In the **Issues** section of the release page, pick the **Unhandled** tab, then click **Open in Issues** and set Sort to `Users` to get the top five crashes by users affected. 5. **Judge**: Healthy when the new rate is within 0.3 points of the previous release, regressed when it is lower by more than that with at least 1,000 sessions, early when adoption is under 10% or under 1,000 sessions. Note: The Releases list rounds crash-free rates to two decimals and the release page time range defaults to the whole life of the release, so set 24H on both to compare like with like. Projects without session data show no health columns at all. ## The report - **Crash-free sessions** for the latest release, and for the previous one over the same window. - **Crash-free users** alongside. - **Adoption**: the share of sessions on the new release, and the session count behind the rate. - **Top crashes** in the release, by users affected. - A verdict: healthy, early or regressed. ## 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": "sentry-crash-free-sessions/", "item_key": "sentry-crash-free-sessions", "project": "acme-ios", "source": "Sentry crash-free sessions for the latest release skill", "kind": "metric", "title": "Release 4.12.0 is 99.62% crash-free", "metric": { "label": "Crash-free sessions", "value": 99.62, "unit": "%", "better": "higher", "period_start": "", "period_end": "" }, "glance": [ { "type": "Grid", "props": { "columns": 3 }, "children": [ { "type": "Stat", "props": { "label": "Previous (4.11.2)", "value": 99.71, "unit": "%" } }, { "type": "Stat", "props": { "label": "Crash-free users", "value": 99.8, "unit": "%" } }, { "type": "Stat", "props": { "label": "Adoption", "value": 38, "unit": "%" } } ] } ], "widget": [ { "type": "Table", "props": { "columns": [ "Top crash in 4.12.0", "Users", "Events" ], "align": [ "left", "right", "right" ], "rows": [ [ "EXC_BAD_ACCESS in ImageCache.prefetch", 61, 74 ], [ "NSInvalidArgumentException in CheckoutViewController", 23, 25 ], [ "SIGABRT in AVCaptureSession startRunning", 9, 11 ] ] } }, { "type": "KeyValue", "props": { "items": [ { "key": "Sessions on 4.12.0", "value": "48,210" }, { "key": "Crashed sessions", "value": "183" } ] } }, { "type": "Callout", "props": { "text": "0.09 points below 4.11.2, inside the 0.3-point margin. ImageCache.prefetch is behind 74 of the 183 crashes.", "tone": "info" } } ], "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: Every 4 hours (`0 */4 * * *`). From cron, with the agent signed in on that machine: ```bash 0 */4 * * * claude -p "Run the sentry-crash-free-sessions 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/sentry-crash-free-sessions#schedule Source: https://briefkin.dev/skills/sentry-crash-free-sessions