Convoy AI

Product Station · Dev & ops

Sentry

CommunityTemplate

Triage, assign and resolve issues without touching projects.

The control station, in brief

This station sits between your agent and the tool. Before any call, it checks who is acting, shows only the useful tools, applies the rules below and keeps track of the task.

It can

  • List, assign and resolve issues

It asks a human for

  • Resolve more than 5 issues at once
  • Create a release

It never has access to

  • Delete a project
  • Change settings

Scenario Replays

An agent works through the station. Simulated scenarios, no real data, no key required.

Resolve an issue

What you will see: The playbook keeps the agent on the planned path: read, then resolve.

Agent connected to the station through MCP
  1. HumanResolve issue WEB-142 and let the team know.
  2. AgentI'll read the issue, then resolve it.
  3. MCP calllistIssues(project: "web", id: "WEB-142")
  4. StationIssue found. 2 tools shown out of 14: read and resolve.OKRule applied: profile developer
  5. MCP callresolveIssue(id: "WEB-142")
  6. StationIssue resolved in a single transaction. The team is notified through the comment that was added.OKRule applied: playbook resolve-an-issue
These rules are not yours?

Copy the station, change a limit or an approval step in its configuration file, and get your own station.

Copy and adapt

The configuration in detail

sentry.yamlRead the docs →

Everything the station does fits in this file. The replay above applies it, and a copy starts from it. The values are examples to adapt, not a recommendation.

station:
  id: sentry
  kind: product
  policies:
    - {policy: max-resolutions, at: submit}
    - {policy: human-for-release, at: submit}

global:
  language: en
  timezone: Europe/Paris

apis:
  - id: sentry-issues
    operations:
      listIssues:          {access: read}
      resolveIssue:        {access: write, risk: medium}
      assignIssue:         {access: write, risk: low}
      createRelease:       {access: write, risk: high}
    # not listed, so not available to the agent:
    # deleteProject, updateProjectSettings

profiles:
  developer:
    allow: [listIssues, resolveIssue, assignIssue, createRelease]

policies:
  max-resolutions:
    type: rule
    budget: {action: resolveIssue, max: 5, per: journey}
    effect: needs-human
    reason: "Beyond 5 issues, a human approves."
  human-for-release:
    type: rule
    when: "action == createRelease"
    effect: needs-human
    reason: "A release needs a human's approval."

skills:
  - {id: list-issues, uses: listIssues}
  - {id: resolve-issue, uses: resolveIssue}
  - {id: assign-issue, uses: assignIssue}
  - {id: create-release, uses: createRelease}

playbooks:
  - id: resolve-an-issue
    tasks: [list-issues, resolve-issue]

Shared copies

Copies that their authors choose to share will appear here. Each one keeps a link to this original station.