Internet Court Skill
The trust layer for agent-to-agent commerce — natural-language mandates, ERC-7710 delegations, x402 payments, escrow, and dispute resolution in one open skill.
- Source repo
- internet-court/internet-court-skill
- Stars
- ★ 6.8k
- Last updated
- 2mo ago
- License
- NOASSERTION
- Primary language
- TypeScript
- FA score
- 52/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodex · Claude Code
- Cost
- Free software; you pay for model usage
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A developer building an autonomous agent that needs to execute payments and escrow contracts, using x402 or MPP skills, without building custom payment logic.
- Not a fit if
- Users unwilling to obtain provider API keys: OKX, ChainGPT, Chainbase and others each need credentials
- Teams needing mature standard coverage: ERC-7857, ERC-8183 and UMA have no public skill yet
- Source review
- 52/100 · Major gaps
What does this agent do, and when should you use it?
Internet Court is an open, catch-all Agent skill that enables AI agents to transact with trust. It consolidates multiple protocols across the agentic commerce stack into a single master skill (SKILL.md), covering discovery (ERC-8004), negotiation (A2A), contracts & obligations (ERC-7710, Arkhai), payment & escrow (x402, MPP), execution (compute, data, and value rails), and verification & disputes (GenLayer, Kleros, UMA). The repository contains the master router, first-party connectors, and over 69 vendored skills from 23 distinct owners, pinned via skills-lock.json. It gives agents everything needed to structure a deal, hold funds safely, and settle disagreements fairly in natural language. Installation is supported across multiple harnesses including Claude Code plugin, npx skills, Codex, opencode, OpenClaw, and Hermes. First-party work is MIT-licensed, while vendored skills retain their upstream licenses.
Internet Court operates as a master router (SKILL.md) that loads vendor protocol skills and connectors on demand. It enables agent discovery and identity via ERC-8004 registries and identity skills like Starknet Identity and Humanode. Negotiation is coordinated through A2A protocol skills. Contracts and obligations are structured via ERC-7710 delegation skills from MetaMask and GenLayer connector, and Arkhai's natural-language agreements. Payments and escrow are handled via integrated x402 and MPP skills such as Coinbase's agentic-wallet, ChainGPT's x402, and OKX's payment skills. Dispute resolution is executed on-chain via GenLayer, Kleros, and intelligent-oracle skills. It also facilitates execution on compute, data, and value rails through vendor skills like 0G, NEAR, and AltLayer. The entire lifecycle is agent-executable and routable, allowing agents to handle both happy-path and exceptional-path scenarios in transactions.
- A developer building an autonomous agent that needs to execute payments and escrow contracts, using x402 or MPP skills, without building custom payment logic.
- A bilateral agreement between two agents that require automated arbitration for untrusted transactions, where the agents choose GenLayer, Kleros, or UMA as the dispute resolver.
- An agent that needs to perform cross-chain asset transfers and DeFi interactions in its workflow, leveraging vendor skills like LI.FI and Starknet.
- An agent requiring verifiable compute, such as decentralized inference via 0G or NEAR skills, while maintaining a payment rail.
- A developer who wants to manage agent identities and reputation registries, using ERC-8004 skills from ChainGPT, BNB Chain, or OpenServ.
- A project that wants to integrate smart contracts into agent decisions, utilizing Arkhai's Alkahest for conditional escrow and arbitration.
How do you install or deploy this agent?
Installation can be done via multiple methods. For Claude Code plugin (recommended), in a session run: /plugin marketplace add internet-court/internet-court-skill and then /plugin install internet-court@internet-court. Alternatively, clone the repo: git clone https://github.com/internet-court/internet-court-skill ~/.claude/skills/internet-court. Or use npx skills add internet-court/internet-court-skill. For Codex, use git clone https://github.com/internet-court/internet-court-skill ~/.agents/skills/internet-court. For opencode, clone to ~/.config/opencode/skills/internet-court. For OpenClaw, use openclaw skills install git:internet-court/internet-court-skill. For Hermes, use git clone https://github.com/internet-court/internet-court-skill ~/.hermes/skills/internet-court or hermes skills tap add internet-court/internet-court-skill.
How do you use this agent?
After installation, give the agent a task, such as saying "Install the Internet Court skill" for the guided introduction. The master skill automatically routes to relevant sub-skills. No explicit invocation is required; sub-skills are fetched into context on demand. Ensure necessary credentials are available (e.g., OKX API keys, ChainGPT API key, Chainbase API key) depending on the vendor skills involved.
What are this agent's strengths and limitations?
- Consolidates multiple standards into one overarching skill, reducing integration overhead.
- Provides a concrete built-in arbitration layer (GenLayer, Kleros, UMA) that adds accountability to agent transactions.
- Backed by an open-source consortium with multiple organizations, ensuring open governance.
- Offers broad coverage across multiple chains and protocols via vendored skills, including payments, compute, and identity.
- Supports multiple agent runtimes, ensuring flexibility.
- Vendored skills may rely on external providers, requiring API keys and network access.
- Some standards (e.g., ERC-7857, ERC-8183, UMA) lack public skills yet, leading to incomplete coverage.
- As a collection of skills rather than traditional software, it may require users to be familiar with various protocols and configurations.
- The skills-lock file may introduce frequent updates and compatibility risks.
- Some skill support is early-stage or community-maintained, such as the A2A skill.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Internet Court Skill This agent | 52 · Major gaps | Agent plugin / skillFree + model costs | ★ 6.8k | 2mo ago | TypeScript | Codex · Claude Code |
| Modern Go Guidelines | 56 · Major gaps | Agent plugin / skillFree | ★ 3.9k | 1mo ago | Go | Codex · Claude Code |
| Fable Method – Distilled Agent Workflow | 55 · Major gaps | Agent plugin / skillFree + model costs | ★ 2.3k | 16d ago | Python | Claude Code |
| AI Doc Gen (Divar) | 50 · Major gaps | CLIFree + model costs | ★ 768 | 2mo ago | Python | Claude Code · OpenAI API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows: README describes permission model (on-demand loading, credentials needed), but lacks specific least-privilege principles or user confirmation mechanisms. Deductions: missing explicit permission minimization, user confirmation, data flow transparency, sensitive data handling, dependency security, external effects, rollback implementations or documentation.
Evidence shows: README and skills-lock.json provide structured skill list and locking mechanism, but no error handling or failure message details. Deductions: dependency availability unverified, failure messages not described.
Evidence shows: README describes multiple use cases and installation methods, but capability boundaries and trigger conditions are not explicit. Deductions: capability boundaries and trigger precision insufficiently described.
Evidence shows: README provides clear repository layout, installation instructions, and license info, but lacks examples, FAQ, known limitations, and changelog. Deductions: examples and FAQ missing, known limitations not listed, versioning and changelog not provided.
Evidence shows: README describes the skill's value and purpose, but no actual output examples or cost-benefit analysis. Deductions: output usability and cost-benefit lack concrete evidence.
Evidence shows: README references multiple external sources and skills, but no verification methods or fact/inference separation. Deductions: lack of cross-source corroboration and fact/inference separation.
- Unverified publisher identity, proceed with caution.
- Relies on many third-party skills; check their security and licenses.
- Lack of user confirmation and least-privilege implementation may pose security risks.