We Built an Agent That Runs Our Entire Sprint While We Sleep

The Problem

Sprint execution is its own full-time job. You pull tickets from Jira, figure out the order, assign them, start implementing, run tests, fix what breaks, commit, push, update the ticket status, write the comment about what you did, move to the next one.

That's not engineering. That's project management with a keyboard. And it crowds out the work that actually moves the product forward.

What We Learned

The sprint workflow follows rules. Known rules, with known tools. Pull tickets in priority order. Run the right agent for each type of work. Run tests after each ticket. Commit. Transition to Done. Push at the end.

The gap wasn't capability — it was the agent to execute them. Once we wrote that agent down, the answer was obvious.

What You Can Do About It

The sprint-runner pulls all To Do tickets assigned in Jira, filters out anything that genuinely needs human judgment (external infrastructure changes, content authoring, OAuth console work), then executes each one using the sub-agent workforce.

For each ticket: it reads the acceptance criteria, launches the right combination of db-agent, api-agent, and frontend-agent, runs the full Playwright test suite, fixes failures up to two times, commits with a clean message, and transitions the ticket to Done with instructions for how to verify the feature in production.

After all tickets are processed, it runs a final build check, pushes, and prints a sprint report.

No pauses. No confirmations. It just runs.

Why It Matters

You start a sprint, go to sleep, and wake up to a report showing what shipped, what's blocked, and what tests passed. The sprint overhead is gone.

Your engineers spend their time on the work only they can do: the architecture decisions, the tricky edge cases, the product conversations. The sprint-runner handles the rest.

At Periscoped, we build for teams that want to move fast without breaking things — and we eat our own cooking.