agent-office 0.6.23 → 0.7.0

package/skills/SKILL-create-coworker.md

@@ -0,0 +1,51 @@
+---
+name: agent-office-create-coworker
+version: 1.0.0
+metadata:
+  description: Create and manage AI coworkers in the office system
+  mutating: true
+---
+
+# Create Coworker Skill
+
+This skill guides agents on creating coworkers (AI agent sessions) in the agent-office system.
+
+## Invariants
+
+- **Always use --dry-run first** when creating coworkers to validate parameters
+- Coworker names should be unique and descriptive (e.g., "Alice", "Bob", "DevAssistant")
+- coworkerType should reflect the role: `assistant`, `developer`, `manager`, `reviewer`, etc.
+- Confirm with user before creating production coworkers
+
+## Best Practices
+
+```bash
+# Validate first
+agent-office --sqlite ./data.db create-coworker \
+  --name "NewCoworker" \
+  --coworker-type "assistant" \
+  --dry-run
+
+# Then execute
+agent-office --sqlite ./data.db create-coworker \
+  --name "NewCoworker" \
+  --coworker-type "assistant"
+```
+
+## JSON Input Format (Agent-First)
+
+```bash
+agent-office --sqlite ./data.db create-coworker \
+  --json '{"name": "NewCoworker", "coworkerType": "assistant"}'
+```
+
+## Required Fields
+
+- `name`: Unique identifier for the coworker
+- `coworkerType`: Role classification
+
+## Related Commands
+
+- `update-coworker` - Modify existing coworker properties
+- `delete-coworker` - Remove coworker and all associated data
+- `list-coworkers` - View all coworkers

package/skills/SKILL-cron-jobs.md

@@ -0,0 +1,106 @@
+---
+name: agent-office-cron-jobs
+version: 1.0.0
+metadata:
+  description: Schedule and manage automated tasks
+  mutating: true
+---
+
+# Cron Job Management Skill
+
+This skill guides agents on creating and managing scheduled cron jobs.
+
+## Invariants
+
+- **Always use request-cron workflow for production** (requires approval)
+- **Always use --dry-run** when creating cron jobs
+- Validate cron schedule expression format before creating
+- Include clear timezone specification
+- Document who to notify on completion
+
+## Two Workflows
+
+### 1. Request-Approval Workflow (Production)
+
+For production cron jobs, use the approval workflow:
+
+```bash
+# Request (creates pending request)
+agent-office --sqlite ./data.db request-cron \
+  --name "Daily Report" \
+  --coworker "Alice" \
+  --schedule "0 9 * * *" \
+  --task "Generate daily summary report" \
+  --notify "Bob, Charlie" \
+  --timezone "America/New_York"
+
+# List pending requests
+agent-office --sqlite ./data.db list-cron-requests
+
+# Approve (requires reviewer)
+agent-office --sqlite ./data.db approve-cron-request \
+  --id 1 \
+  --reviewer "Manager" \
+  --notes "Approved for daily execution"
+```
+
+### 2. Direct Creation (Development/Testing)
+
+For development only:
+
+```bash
+agent-office --sqlite ./data.db create-cron \
+  --name "Test Job" \
+  --coworker "TestBot" \
+  --schedule "*/5 * * * *" \
+  --task "Run test suite" \
+  --notify "Developer" \
+  --timezone "UTC" \
+  --dry-run  # Validate first
+```
+
+## JSON Input Format
+
+```bash
+agent-office --sqlite ./data.db create-cron \
+  --json '{
+    "name": "Daily Report",
+    "coworker": "Alice",
+    "schedule": "0 9 * * *",
+    "task": "Generate daily summary",
+    "notify": "Bob, Charlie",
+    "timezone": "America/New_York"
+  }'
+```
+
+## Cron Schedule Format
+
+Standard cron expression: `min hour day month weekday`
+
+Examples:
+
+- `0 9 * * 1` - Every Monday at 9 AM
+- `0 */6 * * *` - Every 6 hours
+- `0 9 * * *` - Daily at 9 AM
+- `*/5 * * * *` - Every 5 minutes (testing)
+
+## Safety Guidelines
+
+1. **Start with --dry-run** to validate
+2. **Use request-cron for production** (enables approval workflow)
+3. **Test schedules** with frequent intervals first
+4. **Document dependencies** in the task description
+5. **Always specify timezone** explicitly
+
+## Monitoring
+
+```bash
+# Check if cron should run this minute
+agent-office --sqlite ./data.db check-cron-jobs --coworker "Alice"
+
+# Get actionable cron jobs for AI execution
+agent-office --sqlite ./data.db list-active-cron-actions --coworker "Alice"
+
+# View execution history
+agent-office --sqlite ./data.db cron-history --id 1 --limit 10
+```

package/skills/SKILL-send-message.md

@@ -0,0 +1,56 @@
+---
+name: agent-office-send-message
+version: 1.0.0
+metadata:
+  description: Send messages between coworkers
+  mutating: true
+---
+
+# Send Message Skill
+
+This skill guides agents on sending messages between AI coworkers.
+
+## Invariants
+
+- **Always use --dry-run first** for important messages
+- Verify recipient names exist via `list-coworkers` before sending
+- Message body should be clear and actionable
+- Use `--json` for complex messages with formatting
+
+## Best Practices
+
+```bash
+# List coworkers first to verify recipients
+agent-office --sqlite ./data.db list-coworkers --fields name
+
+# Send with validation
+agent-office --sqlite ./data.db send-message \
+  --from "Alice" \
+  --to "Bob" "Charlie" \
+  --body "Task complete. Please review." \
+  --dry-run
+
+# Execute
+agent-office --sqlite ./data.db send-message \
+  --from "Alice" \
+  --to "Bob" "Charlie" \
+  --body "Task complete. Please review."
+```
+
+## JSON Input Format
+
+```bash
+agent-office --sqlite ./data.db send-message \
+  --json '{
+    "from": "Alice",
+    "to": ["Bob", "Charlie"],
+    "body": "Task complete. Please review."
+  }'
+```
+
+## Important Notes
+
+- Recipients must exist as coworkers
+- Multiple recipients can be specified
+- Messages are tracked with read/unread status
+- Use `check-unread-messages` to poll for responses

package/skills/SKILL-task-management.md

@@ -0,0 +1,86 @@
+---
+name: agent-office-task-management
+version: 1.0.0
+metadata:
+  description: Manage tasks in the Kanban board
+  mutating: true
+---
+
+# Task Management Skill
+
+This skill guides agents on managing tasks in the agent-office Kanban system.
+
+## Invariants
+
+- **Always use --dry-run for mutating operations** (add-task, move-task, assign-task, delete-task)
+- **Always use --fields** when listing tasks to limit response size
+- Tasks must move through columns in logical order: idea → approved idea → working on → ready for review → done
+- Check task dependencies before moving to "working on"
+
+## Task Board Columns
+
+1. `idea` - New proposals (default)
+2. `approved idea` - Approved, ready to start
+3. `working on` - In progress
+4. `blocked` - Blocked by dependencies/issues
+5. `ready for review` - Awaiting review
+6. `done` - Completed
+
+## Best Practices
+
+```bash
+# List tasks with field filtering (context window discipline)
+agent-office --sqlite ./data.db list-tasks \
+  --fields id,title,column,assignee
+
+# Add task with validation
+agent-office --sqlite ./data.db add-task \
+  --title "Implement auth" \
+  --column "idea" \
+  --assignee "Alice" \
+  --dry-run
+
+# Move task through workflow
+agent-office --sqlite ./data.db move-task \
+  --id 42 \
+  --column "working on"
+
+# Assign to reviewer
+agent-office --sqlite ./data.db assign-task \
+  --id 42 \
+  --assignee "ReviewerBot"
+```
+
+## JSON Input Format
+
+```bash
+agent-office --sqlite ./data.db add-task \
+  --json '{
+    "title": "Implement auth",
+    "description": "Add JWT authentication",
+    "column": "idea",
+    "assignee": "Alice"
+  }'
+```
+
+## Context Window Discipline
+
+```bash
+# Bad - returns all fields for all tasks
+agent-office --sqlite ./data.db list-tasks
+
+# Good - only return needed fields
+agent-office --sqlite ./data.db list-tasks \
+  --fields id,title,column
+
+# Better - filter then field mask
+agent-office --sqlite ./data.db list-tasks \
+  --column "working on" \
+  --fields id,title,assignee
+```
+
+## Related Commands
+
+- `task-stats` - Overview of tasks by column
+- `task-history` - View task's column transitions
+- `list-task-columns` - Show valid column names