Support engineer and developer, one investigation

Small teamsSaaS team with one support engineer and four developers

Divya in support used Claude Code on the repo to work out why one customer's invoices showed the wrong tax. She found it in an hour. Two days later Arjun picked up the bug ticket and his agent started from zero.

How do I share a support engineer's AI investigation with the developer who gets the bug ticket?

With ArcRoom, the support engineer's agent saves the cause with complete_task: what, which files, which customers. The developer's agent checks find_task before digging, gets that answer with who saved it, confirms the cause still holds, then fixes it. If both look the same day, claim_task tells the second one it's taken. Findings typed into the help desk don't reach the playbook.

Also asked as: Support already found the cause, but the developer's agent started from zero. How do I stop that? · How do I pass a bug investigation from support to engineering? · Can a developer's AI agent see what support's agent found?

The work

Divya's agent saves the cause with complete_task: what, which files, which customers.

ArcRoom keeps
  • Reused answer
  • Claim
  • Saved answer
The next agent gets

find_task → match: reuse, saved by divya (2 days ago)

Step by step

  1. Divya's agent saves the cause with complete_task: what, which files, which customers.
  2. Arjun's agent checks find_task before digging and gets her answer with who saved it.
  3. If both look at it on the same day, claim_task tells the second one it's taken.
  4. Arjun's agent confirms the cause still holds, then fixes it.

What the agent sees

find_task → match: reuse, saved by divya (2 days ago)
"Wrong tax on invoices: customers with a VAT ID added after signup keep the old tax rate; rate is cached on the subscription row."

Compared with today's tools

Ticket notes in Linear or Zendesk

Divya writes her findings on the ticket and Arjun pastes them into his agent.

Setup
none
Ongoing
write up and paste each finding, ~10 min

ArcRoom

Every agent reads the team playbook before it works and writes to it when it finishes. Nothing to update by hand.

Setup
2 min per person (npx arcroom init)
Ongoing
None

The ticket is the right record for the customer; agents don't read it on their own, and ArcRoom knows when the cause went stale.

Where it stops

Divya needs a supported agent and the repo. Findings typed into the help desk don't reach the playbook.

Try it in your project

Works with the agents your team already uses, and any agent over MCP or the HTTP API.

More in small teams