THIS IS REQUIRED READING. Read this playbook FIRST before starting ANY task.
Read playbooks:
systemprompt core playbooks show <playbook_id>
How SystemPrompt Works
The Flow
+------------------------------------------------------------------+
| USER |
| | |
| v |
| +-------------+ |
| | PLAYBOOK | <-- User reads playbook |
| +------+------+ to learn how to |
| | interact with agent |
| v |
| +-------------+ |
| | CLI | <-- User sends task |
| +------+------+ via CLI to agent |
| | |
| v |
| +------------------------------------------------------------+ |
| | AGENT | |
| | | | |
| | +-------------+-------------+ | |
| | v v v | |
| | +--------+ +----------+ +---------+ | |
| | | SKILLS | | MCP | | TOOLS | | |
| | | | | SERVERS | | | | |
| | +--------+ +----------+ +---------+ | |
+------------------------------------------------------------------+
1. Agents Execute Tasks
Agents are AI workers that perform tasks using:
- Skills - Reusable capabilities (e.g., "write blog post", "analyze code")
- MCP Servers - External tools and integrations
- Tools - Built-in functions available to the agent
2. Users Interact via CLI + Playbooks
Users should NOT guess how to interact with agents. Instead:
- Find the agent's playbook - Each agent has a corresponding playbook
- Read the playbook - Learn the correct commands and workflow
- Use the CLI - Send tasks to agents following the playbook
| Agent | Playbook | Purpose |
|---|---|---|
| content | content_blog, content_linkedin, etc. |
Content creation |
| blog | content_blog |
Blog post generation |
| Any agent | cli_agents |
General agent interaction |
IMPORTANT: Interacting with Agents
If you are asked to interact with an agent, you MUST:
-
Find the playbook for that agent:
systemprompt core playbooks list --category content # For content agents systemprompt core playbooks list # List all playbooks -
Read the playbook:
systemprompt core playbooks show <playbook_id> -
Follow the playbook instructions - The playbook contains:
- How to send messages to the agent
- Required parameters and format
- Expected responses
- Error handling
-
Use the CLI to interact (example from playbook):
systemprompt admin agents message <agent_name> -m "Your task here" --blocking
DO NOT improvise agent interactions. The playbook has the correct commands.
3. Edit & Update via CLI
Skills, agent instructions, and playbooks can all be edited using the CLI:
# AGENTS - Edit agent configuration and instructions
systemprompt admin agents list # List all agents
systemprompt admin agents show <name> # View agent config
systemprompt admin agents edit <name> # Edit agent config
systemprompt admin agents create # Create new agent
# SKILLS - Edit skill definitions
systemprompt core skills list # List all skills
systemprompt core skills show <name> # View skill details
systemprompt core skills edit <name> # Edit skill
systemprompt core skills create # Create new skill
systemprompt core skills sync --direction to-db -y # Sync to database
# PLAYBOOKS - Edit playbook instructions
systemprompt core playbooks list # List all playbooks
systemprompt core playbooks show <id> # View playbook
# Edit playbook files directly in: services/playbook/
systemprompt core playbooks sync --direction to-db -y # Sync to database
File Locations
| Type | Location | Sync Command |
|---|---|---|
| Agents | services/agents/*.yaml |
Auto-loaded on restart |
| Skills | services/skills/*.yaml |
systemprompt core skills sync --direction to-db -y |
| Playbooks | services/playbook/**/*.md |
systemprompt core playbooks sync --direction to-db -y |
MANDATORY: Read Before Any Task
Agents MUST follow this workflow for EVERY task:
- Read this playbook first -
systemprompt core playbooks show guide_start - Identify the task domain - Find the relevant playbook below
- Read the domain playbook -
systemprompt core playbooks show <playbook_id> - Follow the playbook steps exactly - Do not improvise
- Verify success - Use the playbook's verification steps
DO NOT skip steps. DO NOT guess commands. ALWAYS use playbooks. ALWAYS fix playbooks immediately if bugs are encountered
Master Playbook Index
Guide (Start Here)
| Playbook ID | Description |
|---|---|
guide_start |
THIS PLAYBOOK - Required first read, links to everything |
guide_playbook |
Playbook authoring standards |
guide_recipes |
Quick recipes for common tasks |
guide_documentation |
Documentation authoring standards |
CLI Operations
| Playbook ID | Description | Use For |
|---|---|---|
cli_session |
Session & authentication | Login, profiles, auth |
cli_agents |
Agent management | List, message, configure agents |
cli_services |
Service lifecycle | Start, stop, restart services |
cli_database |
Database operations | Queries, status, tables |
cli_logs |
Log management & SQL queries | View, stream, trace, direct SQL |
cli_analytics |
Analytics, tracking & bot detection | Stats, costs, traffic, sessions |
cli_cloud |
Cloud setup & management | Auth, tenants, profiles, secrets |
cli_jobs |
Job scheduler | List, run scheduled jobs |
cli_users |
User management | List, roles, permissions |
cli_config |
Configuration | View, update config |
cli_files |
File operations | Upload, list files |
cli_skills |
Skill management | List, sync skills |
cli_contexts |
Context management | Conversation contexts |
cli_plugins |
Plugin management | MCP servers, extensions |
cli_build |
Build system | Compile, test |
cli_deploy |
Deployment | Deploy to environments |
cli_sync |
Data sync | Sync between environments |
cli_web |
Web frontend | Web service config |
Build & Development
| Playbook ID | Description | Use For |
|---|---|---|
build_architecture |
Extension architecture | Understanding the system |
build_extension-checklist |
Extension checklist | Building extensions |
build_extension-review |
Extension review | Reviewing extensions |
build_extension-cli |
CLI extension checklist | Building CLI extensions |
build_mcp-checklist |
MCP server checklist | Building MCP servers |
build_mcp-review |
MCP server review | Reviewing MCP servers |
build_rust-standards |
Rust coding standards | Code style & patterns |
build_crate |
Umbrella crate usage | Using systemprompt crate |
build_web-content |
Web content | Markdown, frontmatter |
build_web-templates |
Web templates | Handlebars, variables |
build_web-assets |
Web assets | CSS, JS, fonts |
Content Creation
| Playbook ID | Description | Use For |
|---|---|---|
content_blog |
Blog posts | Writing blog content |
content_linkedin |
LinkedIn posts | LinkedIn content |
content_twitter |
Twitter/X posts | Twitter content |
content_medium |
Medium articles | Medium content |
content_substack |
Substack posts | Substack content |
content_reddit |
Reddit posts | Reddit content |
content_hackernoon |
HackerNoon articles | HackerNoon content |
How to Use Playbooks
# List all available playbooks
systemprompt core playbooks list
# Read a specific playbook (full content)
systemprompt core playbooks show guide_start
systemprompt core playbooks show cli_session
systemprompt core playbooks show build_architecture
# Raw markdown output (for piping)
systemprompt core playbooks show cli_agents --raw
# Filter list by category
systemprompt core playbooks list --category cli
systemprompt core playbooks list --category build
systemprompt core playbooks list --category content
Task to Playbook Quick Reference
| Task | Read This Playbook |
|---|---|
| Login / authentication | systemprompt core playbooks show cli_session |
| Cloud setup | systemprompt core playbooks show cli_cloud |
| Start/stop services | systemprompt core playbooks show cli_services |
| Send message to agent | systemprompt core playbooks show cli_agents |
| Query database | systemprompt core playbooks show cli_database |
| View logs / SQL queries | systemprompt core playbooks show cli_logs |
| Analytics & tracking | systemprompt core playbooks show cli_analytics |
| Deploy application | systemprompt core playbooks show cli_deploy |
| Create blog post | systemprompt core playbooks show content_blog |
| Build extension | systemprompt core playbooks show build_extension-checklist |
| Build MCP server | systemprompt core playbooks show build_mcp-checklist |
| Understand architecture | systemprompt core playbooks show build_architecture |
Playbook Structure
Every playbook contains:
- Title & Description - What the playbook does
- Prerequisites - Required setup (e.g., active session)
- Step-by-step commands - Exact commands to run
- Verification - How to confirm success
- Troubleshooting - Common issues and fixes
- Quick Reference - Command summary table
Rules for Agents
- ALWAYS read
guide_startfirst - This playbook - ALWAYS read the domain playbook - Before executing any task
- NEVER guess commands - Only use commands from playbooks
- NEVER skip steps - Follow playbooks in order
- ALWAYS verify - Use verification steps from playbooks
- Self-repair playbooks - If a playbook is wrong, fix it before continuing
MANDATORY: Self-Repair
If a playbook command fails or doesn't match CLI behavior, you MUST fix the playbook BEFORE continuing with your task.
When to Self-Repair
- A command in the playbook returns an error
- The CLI syntax has changed
- A command flag doesn't exist
- The playbook references outdated behavior
Self-Repair Workflow
- Stop your current task - Do not continue with broken instructions
- Verify the correct CLI syntax:
systemprompt <domain> <subcommand> --help - Edit the playbook file directly - Playbooks are in
services/playbook/ - Sync the updated playbook:
systemprompt core playbooks sync --direction to-db -y - Verify the fix:
systemprompt core playbooks show <playbook_id> - Resume your task - Now using the corrected playbook
Playbook File Locations
services/playbook/
├── guide/ # guide_* playbooks (onboarding, meta)
├── cli/ # cli_* playbooks (all CLI operations)
├── build/ # build_* playbooks (development standards)
└── content/ # content_* playbooks (content creation)
Example: Fixing a Broken Command
# 1. Command in playbook fails
systemprompt admin agents list --verbose
# Error: unknown flag --verbose
# 2. Check correct syntax
systemprompt admin agents list --help
# 3. Edit the playbook file
# Edit: services/playbook/cli/agents.md
# Change: --verbose -> -v (or remove if not supported)
# 4. Sync changes
systemprompt core playbooks sync --direction to-db -y
# 5. Verify fix
systemprompt core playbooks show cli_agents
# 6. Resume task with corrected command
DO NOT continue with a broken playbook. Fix it first.
Sync Playbooks
If playbooks are missing or outdated:
# Sync from disk to database
systemprompt core playbooks sync --direction to-db -y
# Verify sync
systemprompt core playbooks list