Use casesTriage

Dependency alerts triaged before the week starts

Mitra can triage dependency alerts for you: every Monday at 7am, an agent checks GitHub for new security advisories on the packages you depend on, opens one Linear issue per affected repository, and flags anything critical in Slack. Quiet weeks produce nothing at all.

Runs: Mondays at 7am
Used by: Engineering

The prompt

Every Monday at 7am, check our repositories for new dependency security advisories, open one Linear issue per repository listing the affected packages and their severity, and post only the critical ones to #eng.

GitHub

Linear

Slack

Build this agent

200 credits to start · no credit card

What the agent actually does.

Check for new advisories
The agent reads the security advisories GitHub has raised against your repositories since the last run.
Group them by repository
Every package flagged in one repository becomes a single item, because that is how the upgrade will actually happen.
Open one issue per repository
Each Linear issue lists the affected packages, the severity of each, and the version that resolves it.
Flag the critical ones in Slack
Only critical and high-severity findings reach the channel; everything else waits quietly on the board.

Make it yours.

For teams whose dependency alerts arrive as email nobody reads, and get looked at properly the week after something goes wrong. Change any of this by saying so — there is nothing to rewire.

Run daily instead of weekly for the repositories that ship to production every day.
Include the upgrade command in each issue, so the fix starts from a copy and paste.
Route each repository's issue to the team that owns it rather than one shared backlog.

Questions, answered.

How do I keep track of security advisories without reading every alert?

Mitra collapses the week's advisories into one issue per repository and only interrupts you for the critical ones. Advisory feeds are noisy because every alert arrives at the same volume; the agent sorts by severity before anything reaches Slack. The rest waits in Linear until someone picks the work up.

Why open one issue per repository instead of one per advisory?

Upgrades happen per repository: one person opens one pull request and bumps several packages at once. Mitra groups the advisories to match, so your board shows real tasks instead of thirty near-identical tickets. Ask for one issue per advisory instead and the agent does that.