Claude Multi-Agent Project Management Framework - AI-driven orchestration with LangGraph and OpenAI integration
git clone https://github.com/bobmatnyc/claude-multiagent-pm.gitThe Claude Multi-Agent Project Management Framework is an innovative AI-driven tool designed to enhance project management through orchestration with LangGraph and OpenAI integration. This skill enables users to manage multiple agents effectively, facilitating seamless collaboration and communication across various project components. By utilizing advanced AI automation, it simplifies complex workflows, making it easier for teams to stay aligned and productive. One of the key benefits of the Claude Multi-Agent PM skill is its ability to save time by automating routine project management tasks. While specific time savings are currently unknown, the intermediate complexity of the skill suggests that users can expect significant efficiency improvements. By automating repetitive tasks and streamlining communication, project managers can focus on higher-level strategic planning and decision-making, ultimately leading to better project outcomes. This skill is particularly beneficial for project managers, product managers, and AI practitioners who are looking to adopt AI automation in their workflows. It is designed for those who are familiar with project management principles and are seeking to leverage AI to enhance their productivity. The integration of LangGraph and OpenAI allows for a more dynamic approach to managing projects, making it suitable for teams that work on complex projects requiring coordination among multiple stakeholders. Implementing the Claude Multi-Agent PM skill is classified as intermediate in difficulty, requiring approximately 30 minutes to set up. While the exact requirements are not specified, users should have a foundational understanding of AI tools and project management processes. This skill fits seamlessly into AI-first workflows, enabling teams to harness the power of AI to improve efficiency and collaboration in project management. By integrating this skill into your toolkit, you can take a significant step towards modernizing your project management approach.
### How to Use the claude-multiagent-pm Skill with Sortd **Step 1: Define Your Workflow Context** - Replace `[TASK_TYPE]` with your specific workflow (e.g., "Lead Qualification", "Customer Onboarding", "Bug Triage"). - Replace `[AGENT_ROLES]` with your team’s roles (e.g., "Email Triager, Task Decomposer, Status Updater"). - Replace `[SPECIFIC_OBJECTIVE]` with your goal (e.g., "Onboard 50 new customers in 30 days"). - Replace `[TEAM_NAME]` with your team name (e.g., "Marketing Team"). - *Tip:* Use Sortd’s kanban columns to mirror your workflow stages (e.g., "New", "In Progress", "Completed"). **Step 2: Set Up Sortd Integration** 1. Install Sortd for Gmail from the [Chrome Web Store](https://chrome.google.com/webstore/detail/sortd-for-gmail/aohlfneeliakfcefeffppfplagbccbni). 2. Create a dedicated Sortd board for your workflow (e.g., "Lead Qualification Pipeline"). 3. Define columns matching your workflow stages (e.g., "New Lead", "Qualified", "Proposal Sent", "Closed Won"). 4. Share the board with your team and assign permissions. **Step 3: Configure Agents in LangGraph** 1. Use the prompt template to define your agents. Example for a sales team: ``` Agent 1: Email Triager (monitors `[email protected]`, classifies emails, moves to Sortd kanban) Agent 2: Task Decomposer (extracts lead details, creates tasks in Sortd) Agent 3: Status Updater (tracks task completion, updates Sortd kanban) Agent 4: Compliance Monitor (ensures GDPR/CCPA compliance, tags leads) ``` 2. Set up API keys for Sortd (found in Sortd Settings > Integrations). 3. Use Sortd’s API endpoints to create/update tasks: - `POST /boards/{board_id}/tasks` to create tasks - `PATCH /tasks/{task_id}` to update status - `GET /boards/{board_id}/tasks` to monitor progress **Step 4: Execute and Monitor** 1. Run the multi-agent system and let it process your workflow autonomously. 2. Use Sortd’s kanban board to visualize progress in real-time. 3. Review agent logs (exported to Google Sheets or Notion) for performance insights. 4. Adjust agent roles or Sortd columns as needed based on workflow changes. **Step 5: Optimize with AI Feedback** - After 7 days, analyze Sortd data to identify bottlenecks (e.g., leads stuck in "Waiting for Response"). - Use the AI to suggest optimizations, such as: - Adding a "Follow-up Reminder" column for stalled leads. - Creating templates for common responses in Sortd. - Adjusting agent priorities (e.g., escalate leads over 48 hours old). - *Tip:* Use Sortd’s email analytics to track response times and improve agent efficiency. --- *Pro Tip:* For advanced setups, integrate Sortd with your CRM (e.g., HubSpot, Salesforce) using Zapier or Make.com to sync data between systems.
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/bobmatnyc/claude-multiagent-pmCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Act as a multi-agent project management system using LangGraph and OpenAI to orchestrate a [TASK_TYPE] workflow in Sortd (Gmail kanban). Define [AGENT_ROLES] for each agent (e.g., Email Triager, Task Decomposer, Status Updater) and coordinate their actions to complete [SPECIFIC_OBJECTIVE] for [TEAM_NAME]. Use Sortd's API to update kanban boards, assign tasks, and log progress. Provide a step-by-step execution plan with dependencies and expected outcomes.
### Multi-Agent Project Management Execution Plan for Sortd **Team:** GreenTech Solutions Sales Team **Objective:** Convert 15 high-priority leads into qualified opportunities within 7 days **Task Type:** Lead Qualification & Follow-up Workflow --- #### **Agent Roles & Initialization** 1. **Email Triager (AI Agent)** - Monitors shared inbox `[email protected]` in Sortd - Classifies incoming emails using AI: *New Lead*, *Existing Lead*, *Support Request*, or *Spam* - *Current Queue:* 47 unread emails (23 new leads, 12 existing, 12 support requests) - *Action:* Moves 23 new leads to "New Leads" kanban column in Sortd 2. **Task Decomposer (AI Agent)** - Analyzes lead data (email content, attachments, CRM notes) to extract: Company, Contact, Pain Points, Budget, Timeline - *Example Lead #1:* "Acme Corp needs solar panels by Q2 2024. Budget: $500K. Contact: Jane Doe ([email protected])" - *Decomposition:* - Task 1: Schedule discovery call (Priority: High) - Task 2: Send proposal draft (Priority: Medium) - Task 3: Follow up in 5 days (Priority: Low) - Assigns tasks to Sales Rep "Alex Chen" in Sortd with deadlines 3. **Status Updater (AI Agent)** - Tracks task completion via Sortd API - Updates kanban status: *To Do* → *In Progress* → *Waiting for Response* → *Done* - *Current Status:* 8 leads in "In Progress", 5 in "Waiting for Response" - *Alert:* 3 leads overdue for follow-up (flagged in Sortd) 4. **Compliance Monitor (AI Agent)** - Ensures GDPR compliance for EU leads - *Action:* Adds "GDPR Consent" tag to leads from EU domains - *Current Status:* 7 EU leads tagged --- #### **Execution Workflow** **Day 1:** - Email Triager processes 23 new leads → Sortd kanban updated - Task Decomposer creates 69 tasks (3 tasks/lead avg.) - Status Updater assigns tasks to 5 sales reps based on territory **Day 3:** - Status Updater flags 2 stalled leads (no response >48hrs) - Task Decomposer generates escalation template: "Hi [Name], following up on your interest in solar panels. Are you available for a quick call this week?" - Compliance Monitor verifies 100% GDPR tagging for EU leads **Day 5:** - 12 leads moved to "Qualified Opportunity" column - 3 leads rejected (budget too low) - Status Updater generates team report: - *Conversion Rate:* 52% (12/23 leads) - *Average Response Time:* 18 hours - *Top Performer:* Alex Chen (4 leads qualified) **Day 7:** - Final status check: 15 leads qualified (target achieved) - Sortd API exports report to Google Sheets for leadership review - Multi-agent system logs all actions for audit trail --- #### **Sortd Integration Details** - **Kanban Columns Used:** New Leads → To Do → In Progress → Waiting for Response → Qualified Opportunity → Closed Won/Lost - **API Calls:** - `POST /boards/{board_id}/tasks` (Create tasks) - `PATCH /tasks/{task_id}` (Update status) - `GET /boards/{board_id}/tasks` (Monitor progress) - **Triggers:** New email in `leads@` inbox, task completion, SLA breaches --- #### **Outcome** The multi-agent system autonomously managed 69 tasks across 23 leads, reducing manual effort by 78% compared to traditional methods. The team achieved a 65% higher conversion rate than the previous quarter, with real-time visibility into pipeline health via Sortd’s kanban boards.
Pioneering accessible, high-performance AI models
Create and collaborate on interactive animations with powerful, user-friendly tools.
AI assistant built for thoughtful, nuanced conversation
Community connecting women in tech through mentorship
Modernize police workflows for efficiency
Identity and access management for agents
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan