Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply the Theory of Planned Behavior to predict behavioral intentions from attitudes, subjective norms, and perceived behavioral control, and identify intervention leverage points. Use this skill when the user needs to predict adoption of a new behavior, diagnose why an intended behavior does not occur, design behavior change campaigns, or when they ask 'why do people not follow through', 'what predicts behavior change', or 'how to increase adoption rates'.
.claude/skills/asgard-ai-platform-grad-tpb/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 37% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 66% | 0% |
The Theory of Planned Behavior (Ajzen, 1991) extends the Theory of Reasoned Action by adding perceived behavioral control as a predictor of both intention and behavior. Behavioral intention is determined by three factors: attitude toward the behavior, subjective norms, and perceived behavioral control. Intention is the proximal predictor of behavior, moderated by actual control.
IRON LAW: Intention predicts behavior ONLY when perceived
behavioral control is high — without actual control, intention
alone is insufficient. The intention-behavior gap widens as
volitional control decreases.Key assumptions:
Define the behavior precisely using the TACT framework:
| Predictor | Definition | Underlying Beliefs | |-----------|------------|--------------------| | Attitude | Favorable/unfavorable evaluation of performing the behavior | Behavioral beliefs (outcomes x evaluations) | | Subjective norms | Perceived social pressure to perform or not perform | Normative beliefs (referents x motivation to comply) | | Perceived behavioral control (PBC) | Perceived ease or difficulty of performing the behavior | Control beliefs (facilitators/barriers x power) |
markdown## TPB Analysis: [Target Behavior] ### Behavior Specification (TACT) - Target: [object/person] - Action: [specific behavior] - Context: [situation] - Time: [time frame] ### Predictor Assessment | Predictor | Score | Key Beliefs | Intervention Potential | |-----------|-------|-------------|----------------------| | Attitude | [+/-] | [salient beliefs] | [High/Medium/Low] | | Subjective norms | [+/-] | [key referents] | [High/Medium/Low] | | PBC | [+/-] | [barriers/facilitators] | [High/Medium/Low] | ### Intention Strength: [Strong/Moderate/Weak] ### Intention-Behavior Gap Risk: [High/Medium/Low] ### Recommended Intervention 1. [Primary lever: weakest predictor] 2. [Implementation intention strategy] 3. [Barrier removal or facilitator enhancement]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 18,061 | 24,079 | +33% | 1 | 1 | 0% | 2,748 | 3,196 | +16% | 0 | 0 | — |
case-02 | fail→fail | 28,695 | 17,564 | -39% | 1 | 1 | 0% | 4,486 | 3,834 | -15% | 0 | 0 | — |
case-03 | fail→fail | 28,200 | 14,181 | -50% | 1 | 1 | 0% | 4,978 | 3,381 | -32% | 0 | 0 | — |
case-04 | fail→fail | 22,532 | 17,638 | -22% | 1 | 1 | 0% | 3,623 | 3,809 | +5% | 0 | 0 | — |
case-05 | fail→pass | 17,683 | 8,437 | -52% | 1 | 1 | 0% | 2,675 | 2,397 | -10% | 0 | 0 | — |
case-06 | fail→pass | 15,450 | 11,695 | -24% | 1 | 1 | 0% | 2,299 | 2,987 | +30% | 0 | 0 | — |
case-07 | pass→pass | 20,188 | 13,686 | -32% | 1 | 1 | 0% | 3,141 | 3,229 | +3% | 0 | 0 | — |
case-08 | pass→pass | 18,203 | 24,250 | +33% | 1 | 1 | 0% | 2,744 | 4,778 | +74% | 0 | 0 | — |
case-09 | fail→fail | 20,564 | 18,424 | -10% | 1 | 1 | 0% | 3,260 | 4,139 | +27% | 0 | 0 | — |
case-10 | pass→pass | 16,174 | 12,659 | -22% | 1 | 1 | 0% | 2,623 | 3,230 | +23% | 0 | 0 | — |
case-11 | pass→pass | 16,022 | 12,291 | -23% | 1 | 1 | 0% | 2,401 | 2,988 | +24% | 0 | 0 | — |
case-12 | pass→pass | 16,302 | 11,643 | -29% | 1 | 1 | 0% | 2,565 | 2,976 | +16% | 0 | 0 | — |
case-13 | pass→fail | 19,067 | 15,970 | -16% | 1 | 1 | 0% | 3,138 | 3,957 | +26% | 0 | 0 | — |
case-14 | fail→pass | 21,054 | 13,977 | -34% | 1 | 1 | 0% | 3,518 | 3,043 | -14% | 0 | 0 | — |
case-15 | fail→pass | 15,873 | 15,734 | -1% | 1 | 1 | 0% | 2,520 | 3,461 | +37% | 0 | 0 | — |
case-16 | pass→pass | 40,114 | 16,269 | -59% | 1 | 1 | 0% | 5,901 | 3,585 | -39% | 0 | 0 | — |
case-17 | pass→pass | 25,444 | 18,226 | -28% | 1 | 1 | 0% | 3,836 | 3,506 | -9% | 0 | 0 | — |
case-18 | fail→pass | 15,279 | 19,061 | +25% | 1 | 1 | 0% | 2,469 | 4,108 | +66% | 0 | 0 | — |
case-19 | fail→pass | 24,322 | 12,421 | -49% | 1 | 1 | 0% | 3,475 | 2,953 | -15% | 0 | 0 | — |
case-20 | pass→pass | 15,071 | 15,655 | +4% | 1 | 1 | 0% | 2,469 | 3,614 | +46% | 0 | 0 | — |
case-21 | fail→pass | 41,075 | 15,919 | -61% | 1 | 1 | 0% | 7,290 | 3,682 | -49% | 0 | 0 | — |
case-22 | fail→pass | 34,688 | 13,444 | -61% | 1 | 1 | 0% | 4,519 | 3,130 | -31% | 0 | 0 | — |
case-23 | fail→pass | 24,200 | 13,167 | -46% | 1 | 1 | 0% | 4,075 | 2,959 | -27% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +35 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.