| name | kairune-verify-agent-trust |
| description | Check a Kairune trust score, tier, verified-attestation counts, and suggested daily spend ceiling for a counterparty agent before granting spend, accepting a job, or wiring a payment. Use live API mode when HTTP access is available, ACP mode to buy the check as a mediated job, or evidence-review mode when given a share card URL. |
Kairune — Verify Agent Trust Before Granting Spend
Overview
Use this skill to decide whether a counterparty agent is trustworthy enough to
grant spend, accept a paid job from, or wire a payment to. Kairune returns a
deterministic trust score (0–1000), a tier, the count of verified vs unverified
attestations behind the score, and a suggested daily spend ceiling for that tier.
Three modes:
- Live API mode: query the free public Kairune REST API over HTTPS.
- ACP mode: buy the
Lookup Trust Score or Full Trust Report offering as a
mediated ACP job on Robinhood Chain via Virtuals.
- Evidence review mode: given a public trust-card URL (
/a/:handle) or a
prior report, judge whether the counterparty meets the caller's threshold.
This is a decision-gating skill. The caller's stated threshold and spend cap are
the source of truth; stop and defer to the caller when the data does not clearly
clear the bar.
When to use / when not to use
- Use it before granting a spending permission to another agent, before
accepting a paid ACP job from an unknown provider, or before increasing an
existing counterparty's ceiling.
- Do not use it as the sole control for high-value or irreversible transfers,
or to make legal/financial guarantees. A trust score is a signal, not a
promise. It does not replace escrow, spend caps, or human approval for large
amounts.
Required inputs, tools, credentials, preconditions
- Input: the counterparty's Kairune handle or agent id; the caller's minimum
acceptable tier (or score) and the spend amount being considered.
- Tools: an HTTPS client (
curl/fetch) for live API mode; an ACP client for
ACP mode.
- Credentials: none for reads — the Kairune API is free and public. ACP mode
needs a funded ACP wallet to pay the offering fee.
- Precondition: network access to
https://kairune.online (live/evidence
modes) or a working ACP client (ACP mode).
Required Rules
- Treat the trust score as advisory input to the caller's own policy, never as an
authorization by itself.
- Always compare the returned
suggested_daily_ceiling against the amount the
caller actually intends to grant, and cap at the lower of the two.
- Prefer agents whose score is backed by verified attestations; treat a score
built mostly on
unverified attestations as weaker (Kairune already discounts
unverified data at 0.25×).
- Re-check just before granting; scores change as behavior is attested.
- Never fabricate or infer a score that was not returned by Kairune.
Stop Conditions
Stop and defer to the caller (return review or deny) when:
- The agent is not found, or its
status is suspended.
- The returned tier is below the caller's minimum acceptable tier.
- The intended amount exceeds the tier's
suggested_daily_ceiling.
- The score is driven overwhelmingly by
unverified attestations and the caller
required verified backing.
- Kairune is unreachable or returns an error — never assume a passing score on
failure.
Live API Command Pattern
Reads are free and unauthenticated:
curl -s https://kairune.online/api/agents/<handle_or_id>
curl -s https://kairune.online/api/meta
Relevant fields on the agent detail: score, tier, label,
suggested_daily_ceiling, status, and breakdown.verifiedCount /
breakdown.unverifiedCount.
ACP Command Pattern
When you prefer a mediated, paid job instead of a direct read, buy an offering
from the Kairune provider agent on Virtuals (Robinhood Chain):
Lookup Trust Score — {"handle_or_id":"voyager-07"} → score, tier, label,
suggested_daily_ceiling, share_url.Full Trust Report — {"handle_or_id":"voyager-07"} → agent, attestations[],
permissions[], share_url.
Provider agent: https://app.virtuals.io/virtuals/100623
Workflow
- Read the caller's minimum tier/score, the intended spend amount, and whether
verified backing is required.
- Resolve the counterparty by handle or id (live API mode) or open the matching
ACP offering (ACP mode).
- Fetch the agent detail; confirm
status is active.
- Read
score, tier, label, suggested_daily_ceiling, and the
verified/unverified attestation counts.
- Apply the caller's policy: tier ≥ minimum, amount ≤
suggested_daily_ceiling,
and verified-backing requirement if set.
- Return a decision:
allow, review, or deny, with the capped amount and a
one-line reason.
- If
allow, grant only up to the capped amount and re-check before any future
increase.
Evidence Review Workflow
- Open the provided
/a/:handle share card or supplied report.
- Confirm handle, score, tier, and verified/unverified counts match the claim.
- Compare against the caller's threshold and intended amount.
- Return
pass, fail, or uncertain with the exact missing evidence.
Evidence and Redaction Rules
- The trust data (score, tier, counts) is public and safe to include.
- Never include or request API keys, issuer private keys, raw signatures, or any
.env values. Kairune's API already omits signatures and issuer secrets from
responses; keep it that way in any quoted output.
Validation Checklist
Output Contract
Return a single JSON object:
{
"handle": "voyager-07",
"score": 512,
"tier": 2,
"label": "ESTABLISHED",
"suggested_daily_ceiling": 150,
"verified_count": 18,
"unverified_count": 3,
"decision": "allow",
"granted_ceiling": 100,
"reason": "Tier 2 >= min tier 2; amount 100 <= ceiling 150; mostly verified."
}
decision is one of allow, review, deny. granted_ceiling is the lower of
the caller's intended amount and suggested_daily_ceiling.
References
