Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/openai/skills/llms.txt

Use this file to discover all available pages before exploring further.

The sentry skill gives Codex read-only access to your Sentry workspace through the sentry CLI, letting you investigate production errors, triage unresolved issues, and pull AI-generated root-cause and fix plans — all without leaving your editor. Instead of bouncing between Codex and the Sentry dashboard, you can describe what you are investigating in plain language and have Codex run the right queries and present the findings.

​
Trigger conditions

Activate the sentry skill whenever you want to pull error data from Sentry. Common triggers include:
  • “Show me the top open errors in production from the last 24 hours”
  • “Investigate issue ABC-123 and suggest a fix”
  • “What is the root cause of the latest NullPointerException?”
  • “List unresolved high-priority issues assigned to me”
  • “Summarize recent production errors for this service”

​
What it enables

Issue Listing & Triage

List unresolved issues filtered by environment, time range, priority, and search query. Results include title, short ID, status, first/last seen, count, and top tags.

Issue & Event Detail

Drill into any issue by short ID to view the full detail, then step through individual events including culprit, timestamp, environment, release, and URL.

AI Root-Cause Analysis

Run sentry issue explain to get an AI-generated summary of what caused an issue and the conditions under which it fires.

AI Fix Plans

Run sentry issue plan to receive a prioritized, actionable fix plan for any issue that you can bring directly into your coding session.

​
Core CLI commands

The skill uses the sentry CLI for all queries. The CLI auto-detects your org and project from DSNs in .env files, source code, and directory names — you only need to specify {your-org}/{your-project} if auto-detection picks the wrong target. List the top unresolved issues (last 24 h, production): 
sentry issue list \
  --query "is:unresolved environment:production" \
  --period 24h \
  --limit 20 \
  --json --fields shortId,title,priority,level,status
View an issue by short ID: 
sentry issue view ABC-123 --json
Browse events for an issue: 
sentry issue events ABC-123 --limit 20 --json
AI root-cause analysis: 
sentry issue explain ABC-123
AI fix plan: 
sentry issue plan ABC-123
Arbitrary API access (for endpoints not covered by CLI commands): 
sentry api /api/0/organizations/{your-org}/ --method GET
Use sentry schema <resource> to discover available API endpoints quickly, for example sentry schema issues.

​
Defaults and inputs

ParameterDefaultHow to override
Time range24h--period 7d (or any duration)
Environmentproductionenvironment:staging inside --query
Result limit20--limit 50
Output formatHuman-readable--json for programmatic use

​
Authentication setup

The Sentry CLI must be authenticated before the skill can run. If not yet set up:
1

Install the Sentry CLI

curl https://cli.sentry.dev/install -fsS | bash
2

Authenticate

sentry auth login
Or set the SENTRY_AUTH_TOKEN environment variable directly — never paste your token into the chat.
3

Confirm authentication

sentry auth status

​
Default prompt

When you install the skill, its default prompt is:
“Investigate this issue in read-only Sentry data and report likely root cause, impact, and next steps.”

​
Output rules

The skill follows these formatting conventions for safety and clarity:
  • Issue lists show: title, short ID, status, first seen, last seen, event count, environments, and top tags — ordered by most recent.
  • Event detail includes: culprit, timestamp, environment, release, and URL.
  • PII is redacted from output (emails, IP addresses). Raw stack traces are not printed.
  • Auth tokens are never echoed back.

​
Installing

$skill-installer sentry
Restart Codex after installing to pick up the new skill.

Build docs developers (and LLMs) love

Get started for freeTalk to us