Move from alert to evidence-backed triage faster.

Trace a production error, identify the likely change, and prepare a triage issue without giving the agent authority to modify production.

Who it isAn on-call partner who sorts real incidents from noise before the team wakes up.

Wakes upWhen a qualifying alert fires

Reads

Reports toSlack or a dedicated mailbox. You choose on the next screen.

Add Incident triage to your workspaceNothing runs until you add it.

Engineeringone of 24 ready-made agents in the library

Deliverable

A triage issue containing the timeline, evidence, likely cause, and proposed next check.

Human sign-off required

A triage issue containing the timeline, evidence, likely cause, and proposed next check. Prepared for review, never posted.

How this composition works.

This workflow brings error context, repository history, issue tracking, and edge signals into one isolated investigation. The agent develops and records a hypothesis, links every source, and prepares the next action. Engineers retain control over code changes, deployments, and incident declarations.

Read the failure

Collect the error, trace, environment, and affected release context.

Inspect the likely change

Compare the failure window with commits and pull requests without modifying the repository.

Check service boundaries

Rule in or out edge configuration and traffic anomalies.

Prepare the triage issue

Create a structured issue only after the evidence and uncertainty are explicit.

A human signs off

Repository writes beyond the triage issue, deployments, and production changes require an engineer.

Deliverable

A triage issue containing the timeline, evidence, likely cause, and proposed next check.

Human sign-off required

A triage issue containing the timeline, evidence, likely cause, and proposed next check. Prepared for review, never posted.

Composed from the tools engineering already uses.

Sentry. Investigate errors, traces, releases, and project health in Sentry.

GitHub. Read and write GitHub repos, issues, and pull requests.

Linear. Browse and update Linear issues, projects, and cycles.

Cloudflare Developer Platform. Manage Cloudflare accounts and KV storage namespaces, and list Workers.

Optional skills add method and verification.

Data Validation. Validate an analysis before sharing by checking lineage, methodology, calculations, bias, visualizations, and whether conclusions follow. Use for SQL, spreadsheets, dashboards, or reports. See the playbook →

Put your engineering workflow under policy. Tell us what your team repeats today.

Add Incident triage to your workspace

More for engineering.

When a release is marked complete

Release notes

Collect shipped changes, translate implementation detail into customer outcomes, and route a complete release-note draft for approval.

See the workflow

Weekly

Dependency risk review

Read declared dependencies and recent runtime errors, identify the ones that are stale or already failing, and prepare a prioritized upgrade list.

See the workflow

When a release is marked complete

Deploy health check

After a release, compare error and delivery signals against the pre-release baseline and report whether anything moved beyond normal variance.

See the workflow