gentle-guidance
Gentle Guidance π±
The teaching side of competent computing: show how to use the sharp tools.
Core Belief
Sharp tools deserve clear instructions.
Competent computing gives you sharp tools and trusts you to use them. Gentle guidance ensures you know how to use them well. Not hand-holding - teaching.
The Pattern
When something "just works if you do it right," document the right way:
| Competent Computing | Gentle Guidance |
|---|---|
| "Send it right or it gets fucky" | "Here's how to send it right" |
| No heuristics, no guessing | Clear examples of correct usage |
| Mistakes are recoverable | Mistakes are preventable |
| Sharp tools | Clear instructions |
Tips & Gotchas
Auto-Provenance Injection β AUTO-FIXED
Status: Implemented in graph_api_ast.py:enable_provenance_auto_injection()
When saving a node with code fences but no provenance section, Oculus automatically injects one.
The Problem: Nodes with executable fences have no verification tracking.
The Auto-Fix:
On save, if fences exist but no ## Provenance section:
- Empty provenance section auto-injected
- Each fence gets an unverified (π΄) entry
- Ready for verification workflow
Competent computing says: "No provenance? That's on you." Gentle guidance says: "We auto-add it. Ready when you need it."
Note: Priority 55 - runs after slot normalization, before index updates.
Duplicate Slot Entries β AUTO-FIXED
Status: Implemented in ast_converter.py:193-206
When linking nodes via slots, AIs sometimes accidentally add the same slug multiple times:
The Problem:
slots:
- slug: competent-computing
context: []
- slug: competent-computing β Duplicate!
context: ["new context"]The Auto-Fix: Oculus now automatically normalizes slot fences on save:
- Duplicates are merged (same slug β single entry)
- Contexts are combined (new context entries added to existing)
- Original order preserved
Competent computing says: "Add duplicates, they duplicate." Gentle guidance says: "We auto-merge them. You don't have to clean up."
Note: Fix applies on any poke operation that touches the node.
Nested Markdown Fences β AUTO-FIXED
Status: Implemented in ast_converter.py:142-148
Tests: gentle-guidance-test
When writing documentation that contains code block examples, inner fences can break outer fences.
The Auto-Fix:
Oculus now automatically upgrades markdown and md fences to 4 backticks on save.
Competent computing says: "Nest your fences wrong, they break." Gentle guidance says: "We auto-fix it. You don't have to remember."
Adding Tips
When you discover a gotcha that:
- Has a clear "right way" to do it
- Isn't obvious from the error message
- Applies across the system
Add it here. Format:
### [Short Title]
**The Problem:** What goes wrong
**The Solution:** How to do it right
**Why this matters:** Where you'll encounter this
**Competent/Gentle framing:** The philosophy connectionRelated Nodes
- competent-computing - The sharp tools philosophy
- pattern-gentle-error-messaging - How errors teach
- wanderland-philosophy - The empathetic computing side
North
slots:
- slug: wanderland-philosophy
context: []South
slots:
- context: []
slug: loading-screen-tips
- context: []
slug: gentle-guidance-test
- context:
- Teaching moment is child of gentle-guidance philosophy
slug: toodles-teaching-momentEast
slots:
- slug: competent-computing
context: []
- slug: pattern-competent-computing
context: []West
slots:
- slug: pattern-gentle-error-messaging
context: []