Joan Workflow

This skill should be used when the user asks about "joan", "pods", "workspace", "domain knowledge", "context sync", "joan init", "joan todo", or needs guidance on how Joan's knowledge management system works. Provides workflow guidance for pods, todos, plans, and workspace management.

Security Scan

This is an instruction-only workflow guide for the 'joan' CLI and related workspace concepts; its requirements and instructions are consistent with that purpose and it does not request unrelated credentials or installs.

Purpose & Capability

The name and description say this is a workflow guide for Joan (pods, todos, workspaces). The SKILL.md only tells the agent about CLI commands, local paths (.joan/), and an MCP endpoint — all directly related to that purpose. It does not request unrelated env vars, binaries, or resources.

Instruction Scope

Runtime instructions are limited to using the joan CLI, reading/writing local Joan files (.joan/config.yaml, .joan/pods/, CLAUDE.md), and calling the Joan MCP service. There are no broad, vague directives to harvest arbitrary system files or unrelated environment variables.

Install Mechanism

No install spec or code files are included; this is instruction-only. Nothing will be written to disk by the skill itself during install.

Credentials

The skill declares no required environment variables or primary credential. It mentions that the Joan CLI uses OAuth 2.1 (authenticate via `joan auth login`), which is appropriate and expected for a CLI that talks to a remote MCP server.

Persistence & Privilege

No elevated persistence flags (always, disableModelInvocation) are set. The skill is not force-included and does not request permanent elevated privileges.

Assessment

This appears to be a straightforward usage guide for the joan CLI and is internally consistent. Before installing or relying on it: (1) ensure you install the actual 'joan' CLI from a trusted source (the skill does not provide or install the CLI); (2) be aware that running the CLI will authenticate to joan.land (OAuth) and may upload or download pod content — review what pods contain because they can include sensitive project knowledge; (3) the skill suggests generating CLAUDE.md to inject context into an LLM — avoid including secrets in pods or CLAUDE.md; (4) verify the MCP endpoint (https://joan.land/mcp/joan) is legitimate for your organization. Because this skill is instruction-only, it does not itself exfiltrate data, but running the described commands could send data to the remote Joan service.

Joan is a workspace-based knowledge and task management system for AI-assisted development. This skill covers when and how to use Joan's core concepts.

Core Concepts

Workspaces

Workspaces are the top-level organizational unit in Joan. Each workspace contains:

• Pods: Versioned domain knowledge documents
• Todos: Tasks scoped to the workspace
• Plans: Implementation specs linked to todos
• Members: Team members with roles (admin, member)

Pods

Pods are versioned markdown documents containing domain knowledge. Use pods to:

• Document project architecture and design decisions
• Store domain-specific terminology and business rules
• Share knowledge across team members and AI assistants
• Maintain living documentation that evolves with the project

Pod lifecycle:

1. Create locally with joan pod create
2. Edit the markdown file in .joan/pods/
3. Push to server with joan pod push
4. Pull latest with joan pod pull

Todos

Todos are tasks scoped to a workspace. Use todos to:

• Track work items across team members
• Assign tasks and set priorities
• Link implementation plans to tasks

Todo workflow:

1. Create with joan todo create
2. List with joan todo list
3. Update status as work progresses
4. Archive when complete

Plans

Plans are implementation specs linked to todos. Use plans to:

• Document how a feature will be implemented
• Break down complex tasks into steps
• Share implementation approach with team

CLI Commands Reference

Project Initialization

joan init                    # Interactive workspace selection
joan init -w <workspace-id>  # Non-interactive with specific workspace
joan status                  # Show project and auth status

Pod Management

joan pod list               # List tracked pods
joan pod list --all         # List all workspace pods
joan pod add                # Add workspace pods to project
joan pod create             # Create new pod locally
joan pod pull               # Pull pods from server
joan pod push               # Push local pods to server
joan pod open               # Open pod in browser

Todo Management

joan todo list              # List todos for tracked pods
joan todo list --mine       # List todos assigned to me
joan todo create            # Create new todo
joan todo update <id>       # Update todo fields
joan todo archive <id>      # Archive completed todo

Plan Management

joan plan list <todo-id>    # List plans for a todo
joan plan create <todo-id>  # Create implementation plan
joan plan pull <todo-id>    # Pull plans from server
joan plan push <todo-id>    # Push plans to server

Context Generation

joan context claude         # Generate CLAUDE.md with Joan context

When to Use What

Starting a New Project

1. Run joan init to connect project to a workspace
2. Select pods relevant to the project domain
3. Run joan context claude to inject context into CLAUDE.md
4. Read the generated pod references before coding

Before Coding a Feature

1. Check if relevant pods exist: joan pod list --all
2. Add any missing pods: joan pod add
3. Pull latest: joan pod pull
4. Read pods to understand domain context

After Completing Work

1. Consider if learnings should become a pod
2. Update or create todos to reflect progress
3. Push any local changes: joan pod push and joan todo push

Documenting New Knowledge

1. Create a pod: joan pod create
2. Write domain knowledge in markdown
3. Push to share: joan pod push
4. Update CLAUDE.md context: joan context claude

MCP Integration

Joan provides an MCP server at https://joan.land/mcp/joan with tools:

• list_workspaces - List accessible workspaces
• list_pods - List pods in a workspace
• get_pod - Retrieve pod content

The MCP server uses OAuth 2.1 authentication. Authenticate via the CLI first with joan auth login.

Project Configuration

Joan stores project config in .joan/config.yaml:

workspace_id: <uuid>
tracked_pods:
  - name: "Pod Name"
    id: <uuid>

Pods are stored locally in .joan/pods/ as markdown files.

Best Practices

Pod Authoring

• Use clear, descriptive titles
• Include context about when the knowledge applies
• Keep pods focused on a single domain concept
• Update pods when knowledge evolves
• Reference related pods when helpful

Todo Management

• Create todos at the right granularity (not too big, not too small)
• Link todos to relevant pods for context
• Update status promptly to keep team informed
• Archive completed todos to reduce noise

Context Synchronization

• Run joan context claude after changing tracked pods
• Pull pods before starting significant work
• Push changes promptly to share with team