Jira

All three agents integrate with Jira, but only Buck can be assigned tickets directly. Dolly and Pearl interact with Jira through Slack-initiated workflows — creating tickets, commenting, and reading context — but are not assignable in Jira.

Assigning tickets to Buck

When you assign a Jira ticket to Buck:

1. Reads the ticket description and context
2. Identifies the target repository (from the ticket or by asking)
3. Creates a dashboard session linked to the Jira ticket
4. Dispatches a Claude session in plan mode

Buck focuses on implementation — he turns tickets into code changes.

Dolly and Pearl — Slack-first, Jira-connected

Dolly and Pearl are triggered via Slack (@dolly, @pearl), not Jira assignment. However, they work extensively with Jira:

Dolly

• Creates and updates Jira tickets from Slack conversations
• Decomposes epics into well-defined sub-tickets with acceptance criteria
• Generates status reports and progress summaries from Jira data
• Bulk creates tickets from approved decompositions
• Hands off tickets to Buck (assigns in Jira) or Pearl (adds labels)

Pearl

• Reads Jira tickets for requirements context when tagged in Slack
• Comments design deliverables (reviews, components, UI copy) back on tickets
• Notifies Dolly when design work is complete

Commenting on tickets

Comment on a Jira ticket that Buck is working on to provide follow-up context or instructions. Comments are routed to the same ADK session as the original assignment, maintaining full conversation history.

Note

Each agent filters out its own comments to avoid self-triggering loops.

Session linking

When Buck creates a dashboard session from a Jira ticket, both are linked via the session index. This means:

• Comments on the Jira ticket are seen by the same agent session
• Workflow callbacks (plans completed, PRs opened) are posted back to the Jira ticket
• The agent has full context from both platforms when reasoning about next steps