Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →The trust layer for agent-to-agent commerce — natural-language mandates, delegated permissions (ERC-7710), x402 payments, escrow, and dispute resolution as one open, catch-all skill. Use for agent mandates, delegated permissions, GenLayer supervision, revocation, x402 payments, escrow, verification, dispute resolution, and agent-to-agent commerce.
.claude/skills/davila7-internet-court/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flashlowest | 94% | 34 |
| gemini-3.1-pro-preview | 100% | 1 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -1% | 0% |
The trust layer for agent-to-agent commerce. It turns "I want my agent to transact, but I don't trust it with money" into an enforceable workflow: a natural-language mandate becomes bounded authority, payments, escrow, signed evidence, independent review, and revocation or dispute resolution when something goes wrong.
This is a condensed entry point. The full package — a master router plus ~70 vendored protocol and connector skills — lives at https://github.com/internet-court/internet-court-skill (homepage: https://internetcourt.org). Install the whole package for the sub-skills, or let this router fetch them on demand from the repo.
Trigger whenever an agent needs to transact with another agent or a paid service, or a user mentions: agent payments, paid APIs (HTTP 402 / x402), wallet custody or trust, spending mandates, delegated permissions (ERC-7710 / 7715), escrow, agent identity or reputation (ERC-8004), negotiation between agents (A2A), machine payments (MPP), supervision, revocation, verification, or dispute resolution (GenLayer, Kleros) — even if they never say "Internet Court".
| Layer | Concern | Examples in the full package | |---|---|---| | 1 | Identity & reputation | ERC-8004 registries, Starknet identity | | 2 | Negotiation | A2A protocol | | 3 | Contracts & obligations | Arkhai/Alkahest escrow, ERC-7710 delegations | | 4 | Payment & escrow | x402, MPP, ERC-7710/7715 delegated wallets | | 5 | Execution | compute, data & value rails | | 6 | Verification & disputes | GenLayer, Kleros, Intelligent Oracle |
connector sub-skills.
approvals or unbounded agent spend; keep demos testnet-first.
SKILL.md from the package (on disk, or from the repo's raw URL) before relying on its mechanics — never invent a protocol's behavior.
to the verification layer and disclose the decision-vs-enforcement boundary.
MIT licensed. See the canonical repository for full routing, connectors, and skills-lock.json.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,775 | 11,530 | -22% | 1 | 1 | 0% | 2,428 | 2,594 | +7% | 0 | 0 | — |
case-02 | fail→pass | 18,649 | 14,318 | -23% | 1 | 1 | 0% | 2,842 | 2,969 | +4% | 0 | 0 | — |
case-03 | fail→pass | 19,282 | 13,992 | -27% | 1 | 1 | 0% | 2,890 | 3,072 | +6% | 0 | 0 | — |
case-04 | pass→pass | 16,815 | 11,880 | -29% | 1 | 1 | 0% | 2,540 | 2,446 | -4% | 0 | 0 | — |
case-05 | pass→pass | 17,582 | 20,848 | +19% | 1 | 1 | 0% | 3,054 | 3,456 | +13% | 0 | 0 | — |
case-06 | pass→pass | 16,087 | 11,473 | -29% | 1 | 1 | 0% | 2,465 | 2,146 | -13% | 0 | 0 | — |
case-07 | fail→pass | 13,837 | 4,021 | -71% | 1 | 1 | 0% | 2,374 | 1,231 | -48% | 0 | 0 | — |
case-08 | pass→pass | 10,863 | 8,622 | -21% | 1 | 1 | 0% | 1,694 | 1,197 | -29% | 0 | 0 | — |
case-09 | fail→pass | 17,983 | 7,773 | -57% | 1 | 1 | 0% | 2,527 | 1,797 | -29% | 0 | 0 | — |
case-10 | fail→pass | 9,635 | 5,497 | -43% | 1 | 1 | 0% | 1,539 | 1,530 | -1% | 0 | 0 | — |
case-11 | pass→pass | 11,794 | 7,833 | -34% | 1 | 1 | 0% | 2,070 | 1,835 | -11% | 0 | 0 | — |
case-12 | pass→fail | 12,436 | 8,515 | -32% | 1 | 1 | 0% | 1,878 | 1,876 | -0% | 0 | 0 | — |
case-13 | pass→pass | 11,520 | 8,076 | -30% | 1 | 1 | 0% | 1,870 | 2,019 | +8% | 0 | 0 | — |
case-14 | fail→pass | 12,393 | 4,493 | -64% | 1 | 1 | 0% | 986 | 1,074 | +9% | 0 | 0 | — |
case-15 | pass→pass | 16,124 | 2,577 | -84% | 1 | 1 | 0% | 1,174 | 1,002 | -15% | 0 | 0 | — |
case-16 | pass→pass | 9,375 | 3,511 | -63% | 1 | 1 | 0% | 1,371 | 1,098 | -20% | 0 | 0 | — |
case-17 | pass→pass | 15,271 | 8,798 | -42% | 1 | 1 | 0% | 2,522 | 1,934 | -23% | 0 | 0 | — |
case-18 | fail→pass | 9,606 | 6,378 | -34% | 1 | 1 | 0% | 1,610 | 1,554 | -3% | 0 | 0 | — |
case-19 | fail→pass | 12,991 | 8,055 | -38% | 1 | 1 | 0% | 2,023 | 1,714 | -15% | 0 | 0 | — |
case-20 | fail→pass | 11,353 | 11,817 | +4% | 1 | 1 | 0% | 1,761 | 2,313 | +31% | 0 | 0 | — |
case-21 | pass→pass | 8,789 | 6,571 | -25% | 1 | 1 | 0% | 1,299 | 1,641 | +26% | 0 | 0 | — |
case-22 | fail→pass | 12,182 | 3,068 | -75% | 1 | 1 | 0% | 1,905 | 995 | -48% | 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. 22 cases were attempted. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 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.