Sentry - OpenSRE Documentation

Overview

OpenSRE queries Sentry for recent issues, error events, and stack traces so investigations can correlate application errors with infrastructure alerts. This page covers the REST Sentry integration. For the broader MCP surface (Seer, traces, replays, and more), see Sentry (MCP).

Prerequisites

Sentry account with at least one organization
Auth token with event:read scope (Issues lookup)
For uptime watching: also alerts:read (or org:read) so opensre sentry uptime can list monitors

Setup

Option 1: Interactive CLI

opensre integrations setup sentry

Provide your organization slug and auth token when prompted.

Option 2: Environment variables

SENTRY_ORG_SLUG=your-organization-slug
SENTRY_AUTH_TOKEN=sntrys_your_token
SENTRY_URL=https://sentry.io           # optional, for self-hosted Sentry
SENTRY_PROJECT_SLUG=my-project         # optional, to scope to one project
SENTRY_STATS_PERIOD=24h                # optional, issue search time window

Variable	Default	Description
`SENTRY_ORG_SLUG`	—	Required. Your Sentry organization slug
`SENTRY_AUTH_TOKEN`	—	Required. Auth token with `event:read` (add `alerts:read` for uptime watch)
`SENTRY_URL`	`https://sentry.io`	Override for self-hosted Sentry
`SENTRY_PROJECT_SLUG`	—	Scope queries to a specific project
`SENTRY_STATS_PERIOD`	`24h`	Time window for issue searches (for example `24h`, `14d`, `90d`)

A search returns up to 100 issues per query (Sentry’s maximum page size) within the SENTRY_STATS_PERIOD window. Widen the window (for example SENTRY_STATS_PERIOD=14d) to surface older issues. If you expect more issues than appear, the cause is usually the time window or a SENTRY_PROJECT_SLUG scope — not a hard cap of one page forever.

Option 3: Persistent store

{
  "version": 1,
  "integrations": [
    {
      "id": "sentry-prod",
      "service": "sentry",
      "status": "active",
      "credentials": {
        "base_url": "https://sentry.io",
        "organization_slug": "your-org",
        "auth_token": "sntrys_your_token",
        "project_slug": "my-project"
      }
    }
  ]
}

Credentials

Recommended: Organization Token

In Sentry, go to Settings → Developer Settings → Organization Tokens
Click Create New Token
Enable the event:read scope (and alerts:read if you use uptime watch)
Copy the token

Alternative: Internal Integration For broader access, create an Internal Integration under Settings → Developer Settings → Internal Integrations.

The organization slug appears in your Sentry URL: https://sentry.io/organizations/<slug>/

Investigation tools

Tool	What it does
`search_sentry_issues`	Search issues in the configured org / project window
`get_sentry_issue_details`	Fetch details for one issue
`list_sentry_issue_events`	List error events for an issue
`list_sentry_uptime_alerts`	List uptime monitor health
`get_sentry_uptime_digest`	Roll up recent DOWN / RECOVERED transitions for digests

Verify

opensre integrations verify sentry

Expected output:

Service: sentry
Status: passed
Detail: Sentry validated for org your-org; 30 issue(s) in the last 7 days.

The count reflects issues seen in the last 7 days (capped at 100, shown as 100+ when it saturates). Use search_sentry_issues during an investigation to enumerate issues over a custom window.

Troubleshooting

Symptom	Fix
403 Forbidden	Ensure the token has `event:read`; for uptime watch also `alerts:read`
Organization not found	Verify `SENTRY_ORG_SLUG` matches the slug in your Sentry URL
Connection refused	Check `SENTRY_URL` for self-hosted instances
No issues returned	Normal if none exist in the time window — check `SENTRY_PROJECT_SLUG`
Fewer issues than the Sentry UI shows	Widen `SENTRY_STATS_PERIOD` (for example `14d`) and confirm project scope

Security

Use an Organization Token with only event:read (plus alerts:read if needed) — not an admin token.
Store the token in .env or your secret manager — not in source control.
Rotate tokens periodically.

Extras

Morning digest (scheduled)

Deliver a daily unresolved-issues summary to Slack, Telegram, or Rocket.Chat. This uses the headless sentry-summary skill path (tools + LLM via the gateway), not the investigation pipeline or generic opensre cron kinds. When an uptime watch schedule is running, the skill also calls get_sentry_uptime_digest to include downtime/recovery context.

Command	What it does
`/sentry digest schedule add …`	Same as CLI — schedule from the REPL
`opensre sentry digest run`	Run once and print the digest to stdout
`opensre sentry digest schedule add`	Schedule daily delivery (cron + provider + chat)
`opensre sentry digest schedule list`	List Sentry digest schedules
`opensre sentry digest schedule run TASK_ID`	Run a scheduled digest immediately
`opensre sentry digest schedule remove TASK_ID`	Remove a schedule

Example — weekdays at 08:00 London time to Telegram:

opensre sentry digest schedule add \
  --cron "0 8 * * 1-5" \
  --tz Europe/London \
  --provider telegram \
  --chat-id "-1001234567890"

Example — Slack channel (C… member/channel id):

opensre sentry digest schedule add \
  --cron "0 8 * * 1-5" \
  --tz Europe/London \
  --provider slack \
  --chat-id C0123ABCD

Example — Rocket.Chat (requires token credentials — see Rocket.Chat; an incoming webhook alone cannot target an explicit --chat-id):

opensre sentry digest schedule add \
  --cron "0 8 * * 1-5" \
  --tz Europe/London \
  --provider rocketchat \
  --chat-id "#alerts"

Optional --project my-service scopes the digest to one Sentry project. When an uptime watch schedule is running, the digest includes an Uptime / downtime (last 24h) section. If no uptime watch history exists, the digest stays Issues-only. The gateway daemon picks up scheduled digests when it is running. If the LLM is unavailable, the run fails with an error.

Uptime watch (downtime notifications)

Poll Sentry uptime monitors (Alerts / Uptime — not the Issues digest) and ping Slack, Telegram, or Rocket.Chat only when a monitor transitions to down or recovered. Quiet polls deliver nothing.

Command	What it does
`opensre sentry uptime check`	List current uptime monitor health once
`opensre sentry uptime watch add`	Schedule polling (default every 5 minutes)
`opensre sentry uptime watch list`	List uptime watch schedules
`opensre sentry uptime watch run TASK_ID`	Run a watch tick now (may notify)
`opensre sentry uptime watch remove TASK_ID`	Remove a watch schedule

Example — poll every 5 minutes and ping Slack on transitions:

opensre sentry uptime watch add \
  --cron "*/5 * * * *" \
  --tz UTC \
  --provider slack \
  --chat-id C0123ABCD

Requires alerts:read (or equivalent) on the Sentry token. list_sentry_uptime_alerts exposes the same monitor list in chat/investigation. Schedule an uptime watch alongside the morning digest so sentry-summary can include yesterday’s downtime summary via get_sentry_uptime_digest. Follow-ups (not in v1): inbound Sentry webhooks and auto-remediation attempts.

Telemetry knobs (OpenSRE error reporting)

These control OpenSRE’s own Sentry error reporting — separate from the Issues integration above.

Variable	Default	Description
`OPENSRE_SENTRY_DSN`	—	Override the bundled Sentry DSN
`OPENSRE_SENTRY_DISABLED`	`0`	Set to `1` to disable Sentry entirely
`OPENSRE_SENTRY_LOGGING_DISABLED`	`0`	Set to `1` to disable automatic forwarding of `logger.error` / `logger.exception` without affecting `capture_exception`

Send one test event:

opensre debug sentry

For a custom or self-hosted project:

OPENSRE_SENTRY_DSN=https://public-key@example.ingest.sentry.io/123 opensre debug sentry

Sentry DSN host: example.ingest.sentry.io
Sentry event ID: <event-id>
Sentry flush sent: yes

The event is synthetic and tagged debug=true. If telemetry is disabled, the command exits non-zero without sending anything.

​Overview

​Prerequisites

​Setup

​Option 1: Interactive CLI

​Option 2: Environment variables

​Option 3: Persistent store

​Credentials

​Investigation tools

​Verify

​Troubleshooting

​Security

​Extras

​Morning digest (scheduled)

​Uptime watch (downtime notifications)

​Telemetry knobs (OpenSRE error reporting)

Overview

Prerequisites

Setup

Option 1: Interactive CLI

Option 2: Environment variables

Option 3: Persistent store

Credentials

Investigation tools

Verify

Troubleshooting

Security

Extras

Morning digest (scheduled)

Uptime watch (downtime notifications)

Telemetry knobs (OpenSRE error reporting)