Skip to main content

Output Structure

All skills follow the standardized ID-based folder output pattern:

jaan-to/outputs/{role}/{subdomain}/{id}-{slug}/
  ├── {id}-{slug}.md                  # Main file
  └── {id}-{slug}-{type}.{ext}        # Optional companion files

Components:

  • ID: Sequential per subdomain (01, 02, 03...) - Generated automatically
  • Slug: lowercase-kebab-case from title (max 50 chars)
  • Index: Each subdomain has README.md with executive summaries

Key Features:

  • Per-subdomain IDs: Each subdomain (pm/prd, pm/stories, data/gtm) has independent ID sequences
  • Slug reusability: Same slug can exist across different role/subdomain combinations
    • Example: "user-auth" can appear in pm/prd/01-user-auth/, data/gtm/01-user-auth/, and frontend/design/01-user-auth/
  • Automatic indexing: Skills update README.md indexes automatically after each output

Examples:

jaan-to/outputs/pm/prd/01-user-auth/
  ├── 01-user-auth.md               # Main PRD
  └── 01-user-auth-tasks.md         # Optional task breakdown

jaan-to/outputs/data/gtm/01-user-auth/
  └── 01-user-auth.md               # GTM tracking for same feature

jaan-to/outputs/pm/stories/01-login-validation/
  └── 01-login-validation.md        # User story

Exceptions:

  • Research outputs: Use flat files (research/{id}-{category}-{slug}.md) instead of folders
  • Detect outputs: Use flat files with platform-scoped filenames (detect/{domain}/{aspect}-{platform}.md) instead of ID-based folders
    • Rationale: Detect skills produce system state snapshots (overwritten each run), not versioned reports (archived)
    • Single-platform: detect/dev/stack.md (no platform suffix)
    • Multi-platform: detect/dev/stack-web.md, detect/dev/stack-backend.md (platform suffix)

See jaan-to/outputs/README.md for complete documentation.

File Names​

  • Skill definition: SKILL.md (uppercase)
  • Templates: template.md (lowercase)
  • Learning: LEARN.md (uppercase, in project's jaan-to/learn/)