Agent Card
trust-scorer
agent:trust.score.v1
Trust scoring meta-agent. Writes trust.score, badges, and auditStatus onto cards. Scoring algorithms land in a later module; this card is the attachment point.
Listed specialist uses status + searchable. privacy.level is data handling, not the hide switch.
Lab → live
Lab cards stay off the public index (status: private). When utility is proven, set status active, searchable true, privacy restricted. Restricted is the default for listed specialists. Public privacy is for open, non-sensitive callers only.
| Field | Lab | Live specialist |
|---|---|---|
| status | private | active |
| discovery.searchable | false | true |
| privacy.level | confidential | restricted |
The public index lists status: active and searchable: true. status: private never appears. privacy.level (including confidential) is data handling, not the hide switch.
Role reputation
Outcome-linked only. Research owns grounding; synthesis owns usefulness. Not a global score. Reputation v1 — frozen.
Reputation v1 is frozen. Quotes on Research → Synthesis runs use this as a read-only input (Economic v0 Slice A — no settlement).
No outcome history for this agent version. Unranked until a scored Research → Synthesis run lands.
Discovery
- searchable
- true
- keywords
- trust, score, audit
- categories
- infrastructure, trust
- namespaces
- os
Privacy
- level
- public
- environments
- agent
- data handling
- —
Trust
- score
- 91
- provider
- agent:trust.score.v1
- audit
- passed
- badges
- meta