AgentLedger
Per-agent spend management for AI agents: budget caps enforced before the provider is charged, spend tracking across x402/MPP/API-key rails, anomaly alerts, and audit trails.
Hosted MCP Server
npx add-mcp 'https://aiagentscity.com/mcp/'Installs into Claude Code, Codex, Cursor and more
Documentation
AI Agent City — AI Agent City is the operations layer for the agent economy: spending limits, monitoring, security, trust, and discovery for AI agents. Human-readable and machine-readable.
AI Agent City — agent-readable knowledge index
Umbrella site for seven products, each answering one of two questions — TRUST (verify the counterparty before acting) or OUTCOME (prove the result after): AgentLedger, Agent Watch, Perimeter Watch, TrustScan, Cited, Benefits City, AI Workbench. AgentLedger is the ledger service served by this process; AI Workbench is a local kit installed via npm, not an MCP endpoint.
Machine surfaces
- OpenAPI 3: /openapi.json
- MCP manifest: /server.json · /.well-known/mcp/server-card.json · /.well-known/glama.json
- LLM manifest: /llms.txt
- Buyer skill: /skill.md · /agents.txt · /auth.md
- x402 payment challenge format: /.well-known/x402.json
- Live status: /status
- Sitemap: /sitemap.xml · Robots: /robots.txt
- IndexNow proof: /indexnow.txt
Pricing
- Machine-readable price sheet: /pricing.md
- Human pricing page: /pricing
- Agent-paid rail: 24-hour AgentLedger Pro pass over x402 — see /.well-known/x402.json
Products
- AgentLedger — per-agent spend tracking, budgets, alerts. Landing: /agent-ledger · self-serve workspace: /start [outcome · spine]
- Agent Watch — alerts when new AI agents hit your endpoints: /agent-watch [trust · outward]
- Perimeter Watch — is your endpoint still what it claims: /perimeter-watch [trust · outward]
- TrustScan — npm/MCP package trust scoring: /trust-scan [trust · outward]
- Cited — verify AI-recommendation claims, verbatim evidence: /cited [outcome · outward]
- Benefits City — verified US bank/card signup bonuses, MCP + JSON feed: /benefits/ [outcome · outward]
- AI Workbench — audit trail for agent work; local kit (no MCP): /workbench [trust+outcome · inward]
- Comparison: /compare · All products: /products
Data handling
Cost metadata only — no prompt or response content is collected or stored. See /privacy and /terms.