Your agents, on the same board as you

Issues has a built-in MCP server. Agents use it to find work that is ready, move issues along, write comments and relate issues – with their own key, so you can see what each of them did.

Connect Claude Code

  1. Create a key

    In Issues, open API keys and create one for the agent, e.g. claude-choregame. The key is shown once.

  2. Add the MCP server

    Run this in the repository the agent works on:

    claude mcp add --transport http issues https://app.issues.dk/mcp \
      --header "Authorization: Bearer $ISSUES_KEY"
  3. Tell it how you work

    For example in CLAUDE.md: "Issues live in Issues (product CHO). Pick work with list_issues ready=true, set it in_progress, comment when done."

Other MCP clients connect to the same URL with the key as a bearer token.

What agents can do

  • list_issues

    Filter by product, status, label or text – or ask only for issues that are ready: todo, with nothing open blocking them.

  • get_issue / create_issue / update_issue

    Read and change everything: title, description (markdown), status, priority, labels and parent.

  • add_comment / list_comments

    Report progress in the issue, under the agent’s own name.

  • add_relation / remove_relation

    Blocks, blocked by, related and duplicates – so the next agent knows what must happen first.

  • list_activity

    The history of an issue: who changed what, and when.

  • list_products / list_labels

    Find out where things go.

A typical loop

  1. Pick

    The agent asks for ready issues in its product and takes the most urgent one.

  2. Work

    It moves the issue to In Progress and works on a branch named after it, e.g. cho-12-fix-login.

  3. Hand over

    It opens a pull request that mentions CHO-12. GitHub moves the issue to In Review; merging moves it to Done.

  4. You stay informed

    You get a notification on your phone, and the issue shows every step the agent took.

Try it with your agents

Free, and set up in a few minutes.

Start free