trust-formationlisted
Install: claude install-skill Trustbaselab/trustbase-skills
# Trust Cannot Be Claimed — Only Verified
Based on the ideas of *Who Can You Trust?* (Rachel Botsman, 2017).
## When to use
- "Why should anyone trust this" · "how do we prove we're credible" · "how do we make people believe us" · "how do we write a trust endorsement" · "how do we prove we're trustworthy"
- "Is this source/platform/tool trustworthy" · "institutions aren't trusted anymore, what now" · "you can't trust anybody"
- Before designing trust signals (ratings / certifications / badges / endorsement copy), run a four-element audit first
- **Lightweight mode**: if simply asked "is X trustworthy" → go straight to Step 0 and output the "trust check card" (template 2), without the full pipeline
**Not for**:
- Data provenance / year / definition checks → use `factfulness`
- AI conclusions that are overconfident, or bias review → use `thinking-fast-and-slow`
- "How do you know you actually understand this" calibration → use `knowledge-illusion`
- Attention / multitasking / scheduling problems → use `bandwidth-and-deep-work`
- Internal organizational management and culture building (the part this book does not address) → do not apply this skill
## Three iron rules
### Rule 1: Trust ≠ confidence — trust faces the unknown and cannot be obtained by claiming it
**Trust** is a confident relationship with the unknown, requiring vulnerability and choice; **confidence** comes from past predictability. So the sentence "we are trustworthy" is itself untrustworthy — only testable