Core Concepts
Six building blocks of jaan.to.
Skills
Commands that generate outputs.
- Pattern:
/role-domain-action - Example:
/jaan-to:pm-prd-write - Output: Markdown files in
jaan-to/outputs/
Skills follow a two-phase workflow:
- Analysis - Read context, ask questions
- Generation - Create artifact, preview, write
Stacks
Your team and tech context.
| File | Contains |
|---|---|
jaan-to/context/tech.md | Languages, frameworks, tools |
jaan-to/context/team.md | Team size, ceremonies, norms |
jaan-to/context/integrations.md | Jira, GitLab, Slack setup |
Skills read context to generate context-aware outputs.
Templates
Output formats for outputs.
- Location:
jaan-to/templates/{skill}.template.md - Contains: Section structure, placeholders
- Override: Create your own version
Skills fill templates with generated content.
Learning
System improves from feedback.
Each skill has a LEARN.md file with four categories:
| Category | Example |
|---|---|
| Better Questions | "Ask about i18n requirements" |
| Edge Cases | "Check for multi-tenant scenarios" |
| Workflow | "Generate metrics JSON alongside PRD" |
| Common Mistakes | "Don't assume single region" |
Add lessons: /jaan-to:learn-add {skill} "lesson text"
Hooks
Automated triggers before/after actions.
| Type | When | Example |
|---|---|---|
| PreToolUse | Before write | Validate required sections |
| PostToolUse | After write | Prompt for feedback |
Hooks run automatically. No action needed.
Guardrails
Safety boundaries.
- Safe paths: Only write to
jaan-to/ - Preview first: Always show before saving
- Approval required: You confirm every write
Guardrails cannot be disabled.