Claude Code builds a Pylon support triage dashboard
Strivens reports building StackDash with Claude Code and live MCP connections, producing prioritised Pylon issues, response drafts and deduplicated explicit meeting actions for daily support triage.
stackone.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Build [support triage dashboard] using [Pylon and Fireflies tools]. Validate and deduplicate incoming issues, retain full conversation evidence, draft reviewable replies, and import only explicitly assigned meeting actions with stable source IDs.
Reach people exploring AI agents
See placements →What happened
Strivens reports building StackDash with Claude Code and live MCP connections, producing prioritised Pylon issues, response drafts and deduplicated explicit meeting actions for daily support triage.
Reported by Max StrivensHow the agent did it
- Connect Pylon and Fireflies tools through StackOne MCP.
- Build the dashboard, webhook validation and persistent issue records.
- Retrieve full issue context for priority and investigation briefs.
Where a person stepped in
Strivens directs architecture and prompt quality and checks the dashboard twice daily. Illustrated customer data is an example, and no independent resolution metric is supplied.
Max Strivens builds an internal support dashboard with Claude Code and StackOne MCP connections to Pylon and Fireflies. The agent implements the interface and backend while Strivens directs architecture and iterates on the communication prompts.
Pylon webhooks feed issues into a validated, duplicate-checked pipeline. Full issue details, account information and message threads provide the context for prioritisation and investigation prompts. Fireflies import retrieves meeting summaries by date and participant, retaining only explicitly assigned actions. Transcript and action identifiers prevent repeated imports from creating duplicate tasks.
The deployed dashboard also prepares concise response drafts and code-investigation briefs. Strivens reports checking it twice daily instead of continuously monitoring Slack. The illustrated Acme sync problem is an example format, not evidence of a specific real customer incident, and broader automatic Slack summaries and ticket creation are future work. The useful boundary is a concrete support artifact with source-rich prompts and explicit-only action extraction. Claude Code builds this workflow through MCP tools; the resulting dashboard's runtime classification should not be confused with proof that every customer answer was autonomously investigated or sent. The source gives no independently measured resolution improvement.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · Max Strivens
Frequently asked questions
How are repeated meeting imports handled?
A source identifier combines the transcript identifier and action slug so reimporting does not duplicate the task.
Which meeting actions are extracted?
The prompt requires explicit assignments; inferred owners or tasks are excluded.
Are automatic Slack summaries already deployed?
The source lists webhook-triggered Slack summaries and broader automatic ticket creation as potential future work.
More details
Remaining steps
- Import explicitly assigned meeting actions with source-based deduplication.
- Have Strivens inspect the dashboard and review response drafts.
About this account
Claude Code builds and iterates on the operational dashboard while calling live MCP tools to discover and connect support and meeting data.
Firsthand account by Max Strivens; original publication date verified against Primary StackOne byline March 17, 2026. Outcomes are the author’s reports, not independently reproduced. Suggested prompt written by Agents In Use; not the author’s original prompt.
Credit: Max Strivens
Source published 17 Mar 2026
Source checked 14 Oct 2026
Tools and topics
Claude Code · StackOne MCP · Pylon · Fireflies · React · Cloudflare Workers · Cloudflare KV
Support triage dashboard · Explicit action extraction · Import deduplication
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.