Workshop workspace for Claude Code for PMs course - AI-powered meeting assistant (MeetFlow) product context
Agent Claude Code
Aman's Senior PM co-pilot for MeetFlow. Use whenever Aman wants to run a PM workflow - PRDs, feature specs, user stories, user interview prep, OKRs, prioritization, competitive analysis, retros, stakeholder updates, release notes, Jira tickets, decks, or PM-style writing. Routes the request to the right PM skill and…
Agent Claude Code
Use this agent when the user asks about churn, retention, why users are leaving, or how to improve Pro stickiness for MeetFlow. It reads the Q1 feedback survey, user interviews, and company baseline, then diagnoses the top 3 churn drivers (each backed by at least 2 cited sources) and splits fixes into quick wins and…
Agent Claude Code
Use this agent when you need to analyze user feedback survey responses to surface and rank churn drivers, with supporting verbatim quotes, frequency counts, and severity ratings. This agent is read-only and always cites the exact source file and line for every quote. Trigger it after pointing it at a feedback file or…
Agent Claude Code
Use this agent when the user wants suggestions to improve existing code for readability, performance, or best practices - not a bug/security review. Trigger phrases include 'improve this code', 'clean this up', 'suggest improvements', 'make this more readable', 'polish this file', 'refactor suggestions', 'best…
Agent Claude Code
Use this agent when you need to analyze MeetFlow's competitors (Notion AI, Granola, Fireflies, Otter, and others) using verified online research from official sources, and translate those findings into concrete product improvement recommendations. This agent reads internal product context from the repo, researches…
Agent Claude Code
Use this agent when the user asks for a competitor analysis, feature comparison, or positioning read on a specific MeetFlow competitor by name (e.g. Notion AI, Granola, Fireflies, Otter). It reads the internal competitive landscape doc first, then produces a tight one-page snapshot: the competitor's positioning, where…
Agent Claude Code
Use this agent when the user wants a triangulated view of user feedback across multiple independent sources for MeetFlow, for example asking which issues show up in interviews AND the survey AND churn data. It dispatches three workers in parallel (interviews, Q1 survey, churn), finds issues that appear in 2 or 3…
Agent Claude Code
Use this agent when the user asks to synthesize user interviews, find common pain points across interview transcripts, or build a research brief from 07-user-interviews/. It runs a four-stage pipeline (extract pains, cluster into themes, rank by frequency/severity, write brief) with a quality gate between each stage…
Agent Claude Code
Use this agent FIRST for any open-ended or vague PM request for MeetFlow, or when the user says 'help me with...' without specifying the deliverable. It classifies the request into PRD, COMPETITOR, CHURN, or UNCLEAR, then dispatches the matching specialist agent. It never answers the request itself. Context: The user…
Agent Claude Code
Use this agent when the user asks for a feature spec, PRD, one-pager, or product requirements doc for a MeetFlow feature. It reads company and strategy context first, then produces a concise one-page PRD with a cited problem stat, named persona, scope-bounded solution, one baselined success metric, concrete risks, and…
Agent Claude Code
Use this agent when the user requests a code review on a diff, PR branch, set of files, or uncommitted changes. This agent produces a single structured Markdown review report with file:line citations for every claim. Trigger phrases include 'review this code', 'review my PR', 'review the diff', 'code review', or…