rogee/superpowers
1
0
Fork 0
mirror of https://github.com/obra/superpowers.git synced 2026-04-24 10:29:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add persuasion research foundation and enforce TDD for skill edits

Browse Source 
Based on Meincke et al. (2025) research showing LLMs respond to same persuasion
principles as humans. Created reference doc explaining psychology behind effective
skill design (authority, commitment, scarcity, social proof, unity).

Extended Iron Law to explicitly cover skill edits - agents were skipping
RED-GREEN-REFACTOR when editing existing skills ("just a simple addition").

Changes:
- Add persuasion-principles.md reference doc with research citations
- Update creating-skills when_to_use to include editing skills
- Expand Iron Law with "Edit skill without testing? Same violation"
- Add explicit no-exceptions for "simple additions", "documentation updates"
- Add psychology note reference in bulletproofing section
- Add persuasion psychology section to testing-skills-with-subagents

Baseline testing showed agents already apply persuasion principles intuitively
when creating skills. This change makes the WHY explicit and prevents
rationalization that edits don't need testing.
This commit is contained in:
Jesse Vincent
2025-10-09 22:10:59 -07:00
parent a5d52463f8
commit e3e02fe67d
3 changed files with 201 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
skills/meta/creating-skills/SKILL.md
View File

@@ -1,8 +1,8 @@
---
name: Creating Skills
description: TDD for process documentation - test with subagents before writing, iterate until bulletproof
when_to_use: When you discover a technique, pattern, or tool worth documenting for reuse. When you've written a skill and need to verify it works before deploying.
version: 4.0.0
when_to_use: When you discover a technique, pattern, or tool worth documenting for reuse. When editing existing skills. When asked to modify skill documentation. When you've written a skill and need to verify it works before deploying.
version: 4.1.0
languages: all
---
@@ -317,12 +317,17 @@ When: Reference material too large for inline
NO SKILL WITHOUT A FAILING TEST FIRST
```
This applies to NEW skills AND EDITS to existing skills.
Write skill before testing? Delete it. Start over.
Edit skill without testing? Same violation.
**No exceptions:**
- Don't keep it as "reference"
- Don't "adapt" it while running tests
- Don't look at it
- Not for "simple additions"
- Not for "just adding a section"
- Not for "documentation updates"
- Don't keep untested changes as "reference"
- Don't "adapt" while running tests
- Delete means delete
See skills/testing/test-driven-development for why this matters. Same principles apply to documentation.
@@ -395,6 +400,8 @@ Different skill types need different test approaches:
Skills that enforce discipline (like TDD) need to resist rationalization. Agents are smart and will find loopholes when under pressure.
**Psychology note:** Understanding WHY persuasion techniques work helps you apply them systematically. See persuasion-principles.md for research foundation (Cialdini, 2021; Meincke et al., 2025) on authority, commitment, scarcity, social proof, and unity principles.
### Close Every Loophole Explicitly
Don't just state the rule - forbid specific workarounds: