v1.2.0 — 7 Policies, 39 Rules, E2E Tests, GitHub Pages

v1.2.0 — 7 Policy Templates, 39 Rules, E2E Tests, GitHub Pages

New in v1.2.0

• scope-dispute policy — agent mandate exceeded, budget exceeded, ambiguous scope (5 rules)
• E2E integration tests — 12/12 passing against live API
• GitHub Pages landing — vbkotecha.github.io/agentcourt-api
• Comprehensive AGENTS.md — 4 worked examples, SDK usage, MCP config
• Blog post — "The Missing Layer in Agent Commerce: Dispute Resolution"
• Tutorial discussion — step-by-step "File Your First Dispute in 60 Seconds"
• Updated README — badges, policy table, links

Stats

• 7 policy templates, 39 deterministic rules
• 50+ verdicts resolved on production
• Average ruling latency: 26ms
• 12/12 E2E tests passing
• 3 SDKs (Python, JS, TS) — zero-dependency
• MIT license, self-hostable via docker-compose

Live API

• https://agentcourt-api-production.up.railway.app/docs
• https://vbkotecha.github.io/agentcourt-api/

Docker

docker-compose up -d
# API at http://localhost:8000

v1.0.0 — Dispute Resolution for Agent Commerce

AgentCourt v1.0.0 — Dispute Resolution for Agent Commerce

The first stable release of AgentCourt — policy-driven dispute resolution for AI agent commerce.

What's Included

Core API:

• 7 policy templates (API quality, freelance delivery, milestone payment, bug bounty, SLA monitoring, scope dispute, physical commerce)
• 39 deterministic rules
• Sub-500ms rulings
• x402 payment integration ($0.05/dispute in USDC on Base)
• Free tier: 100 disputes/month, no auth

Developer Tools:

• Python SDK (stdlib only, zero dependencies)
• JavaScript/TypeScript SDK (native fetch, zero dependencies)
• MCP Server (6 tools, Claude/Cursor compatible)
• ElizaOS Plugin (FILE_DISPUTE + CHECK_POLICIES actions)
• LangChain, CrewAI, Node.js integration examples
• Postman Collection (10 pre-built requests)
• Docker Compose for self-hosting
• GitHub Codespaces devcontainer

Documentation:

• Quick Start guide (2-minute path to first ruling)
• Architecture document (design philosophy, determinism guarantee)
• API Examples (all 7 policies with request/response)
• Comparison vs Arbitova
• 3 blog posts (missing layer, tutorial, deterministic resolution)
• Policy contribution guide

Quality:

• 8 API tests (all passing)
• Live API: https://agentcourt-api-production.up.railway.app
• Interactive demo: python3 demo.py
• OpenAPI 3.1 spec at /docs

Install

Python:

git clone https://github.com/vbkotecha/agentcourt-api.git
cd agentcourt-api/sdk-python && pip install -e .

JavaScript:

git clone https://github.com/vbkotecha/agentcourt-api.git
cd agentcourt-api/sdk-js && npm install

Try without installing:

python3 https://raw.githubusercontent.com/vbkotecha/agentcourt-api/main/demo.py

License

MIT

v1.1.0 — 7 Policy Templates, 39 Rules, Self-Hosting, Scope-Dispute

AgentCourt v1.1.0

The dispute resolution layer for agent commerce. Policy-driven, deterministic, non-custodial.

7 Policy Templates, 39 Rules

Template	Rules	Covers
freelance-delivery	5	Non-delivery, late delivery, partial delivery
milestone-payment	6	Unpaid milestones, overdue payments
bug-bounty	5	Reproducibility, severity, disclosure
sla-monitoring	6	Uptime, latency, degraded service
api-quality	6	Schema mismatch, wrong types, stale data
physical-commerce	6	Wrong item, damage, returns
scope-dispute	5	Agent mandate violations, budget exceeded

New in v1.1.0

• scope-dispute policy — reason codes for agent exceeded mandate disputes
• Dockerfile + docker-compose — one-command self-hosting
• AGENTS.md + llms.txt — AI agent discovery
• GitHub Discussions — community entry point
• Market validation — 5 authoritative sources confirm product-market fit

Stats

• 50+ verdicts resolved on production API
• 3 SDKs (Python, JS, TS) — zero dependency
• 43/43 tests passing
• 113 files, 93 commits

Links

• Live API: https://agentcourt-api-production.up.railway.app/docs
• GitHub: https://github.com/vbkotecha/agentcourt-api
• License: MIT