Monthly product feedback synthesis to Jira and Slack

Build me a monthly voice-of-customer synthesis workflow that turns social feedback about our product into Jira tickets the PM team can triage.

Trigger: a cron schedule that runs at 8am local time on the 1st of every month.

On each run, the agent should:

1) Query Infegy Starscape for posts from the trailing 30 days that mention our product paired with feature-request or complaint language (configurable search terms, with a default seed list the agent can edit). Use Get Topics Analysis to cluster the matching conversations into theme groups with size and sentiment.

2) For each cluster above a configurable mention threshold (default 25 mentions), use Download Post Records to pull a representative sample of posts for that theme, then call Generate AI Summary to write a tight description that covers the underlying user pain, who is asking (audience or persona signal from the post records), frequency relative to the prior month, and 2 or 3 example quotes.

3) For every qualifying cluster, use Jira Create Issue in the configured product backlog project. The issue summary is the theme name. The description body contains the synthesized write-up plus the example quotes. Apply a 'voice-of-customer' label and the configured issue type (default Task). If a Jira filter or epic is configured, link the issue to it.

4) After all tickets are filed, use Slack Send a Message to post a digest in the configured product channel. The digest should include the total number of tickets filed this run, the top 3 themes by mention volume with a one-line summary each, the run window, and a link to the Jira filter view that shows all tickets with the voice-of-customer label.

Configurable inputs the workflow should expose: Infegy search terms and brand/product scope, mention threshold per cluster, Jira project key, Jira issue type, Jira label, optional Jira epic or filter URL, Slack channel, and the run timezone. If a month produces zero qualifying clusters, still post a Slack note saying so instead of going silent.

Keep ticket bodies skimmable: short paragraph for the pain, a 'who is asking' line, a 'month-over-month' line, and a quotes section with at most 3 quotes and their sources.