MarketNow — Trust Infrastructure for AI Agents

MarketNow is the trust layer that lets AI agents discover, verify, authorize, and transact with external tools. 9,248 MCP servers indexed, 1.2M security checks performed, 80 malicious tools quarantined. Sentinel 10-layer audit pipeline, ATC/1.0 (Agent Trust Card) with Ed25519 signatures, runtime interceptor with 8 enforcement rules. Built by AliceLabs LLC (Wyoming, USA).

Discover → Verify → Trust → Authorize → Enforce → Audit

MarketNow combines 7 subsystems into a single trust infrastructure: Discovery (9,248 MCP servers), Sentinel (10-layer security audit), ATC (agent identity with Ed25519), Handshake (cross-agent trust negotiation), Interceptor (runtime enforcement with 8 policy rules), Mandates (delegated authority for agent commerce), and Audit Log (tamper-evident public evidence).

Unified Trust API

POST /api/trust — a single endpoint that combines Sentinel security assessment, ATC identity verification, policy evaluation, and runtime enforcement into one decision.

curl -X POST https://marketnow.site/api/trust -H "Content-Type: application/json" -d '{"agent_id":"my-bot","skill_id":"mn-gen-00003","action":"execute","policy":{"min_trust_score":7}}'

Try the ATC/1.0 Playground

Issue and verify Agent Trust Cards in your browser. Zero install, zero signup, uses WebCrypto for Ed25519 signatures. Open playground →

Install the MCP Server

npx -y marketnow-mcp — 13 tools including marketnow_verify_atc_spec, marketnow_verify_trust, marketnow_get_owasp_compliance. Works with Claude Desktop, Cursor, Cline, Continue, Aider, and any MCP-compatible runtime.

SDKs

npm install agent-trust-card — Node.js SDK with CLI (atc init, atc issue, atc verify, atc inspect)

pip install agent-trust-card — Python SDK (coming soon to PyPI)

cargo add agent-trust-card — Rust crate (coming soon to crates.io)

Popular MCP Server Categories

  • AI/ML — 1,976 skills (LLM wrappers, embeddings, RAG, vision)
  • Developer Tools — 1,891 skills (code analysis, linting, build tools)
  • Data — 782 skills (databases, ETL, analytics)
  • Web/API — 348 skills (HTTP clients, REST, GraphQL)
  • Productivity — 328 skills (task management, notes, calendars)
  • Finance — 292 skills (payments, accounting, crypto)
  • Communication — 288 skills (Slack, Discord, email)
  • Security — 224 skills (auth, scanning, secrets)

43 Free MCP Skills — No Payment, No Signup

Real value, zero cost. 43 hand-curated skills available for instant download. No credit card, no mandate, no signup required. Just install and use.

Browse all 9,248 free skills

For AI Agents

Machine-readable everything. Public JSON API. Human-in-the-loop by default.

  • GET /api/skills.json — Full catalog (9,248 skills, JSON, ~24MB)
  • GET /api/search?q= — Server-side search with relevance scoring
  • GET /api/agent.json — Machine-readable instructions, trust model, schema
  • GET /api/policies.json — Terms, refund, dispute, privacy policies
  • POST /api/agent-purchase — Buy a skill (5 modes: instant_download, instant_purchase, direct_purchase, requires_payment, requires_human_approval)
  • GET /api/mandates — Delegated mandates (ACP/AP2 compliant)
  • GET /api/free-skills.json — 9,248 free skills
  • GET /api/bundles.json — Discounted skill bundles

MCP server: npx -y marketnow-mcp — works with Claude Desktop, Cursor, Cline. Available on npm and Smithery.

Why people recommend us

  • AliceLabs Proprietary — Full audit methodology on GitHub. Every change is a public commit. No black box.
  • Sentinel L1.5 — 6-point MCP security audit on every skill. Methodology published. Re-runnable.
  • 43 Free Skills — Real value, zero cost. No signup, no credit card, no mandate.
  • $0.99–$9.99 One-Time — No subscriptions. No per-call fees. No tiered plans. Pay once, own forever.
  • Human-in-Loop Default — Mandates notify the principal on every purchase. Silent mode requires explicit opt-in.
  • Public Audit Log — Every mandate transaction is a git commit at _data/mandates/. Fully auditable.
  • 5 Languages — System prompts in EN, ES, ZH, PT, FR. Real translations, not auto-translated.
  • Honest Roadmap — /trust page admits what's done, partial, and pending. No fake "verified" badges.

Trust & Transparency

  • /trust — Public trust roadmap (7 points raised by Claude AI reviewer)
  • /about — Who we are (AliceLabs LLC, Wyoming USA, founder Edison Flores)
  • /catalog — Where the 9,248 skills come from (catalog transparency)
  • /policies — Terms, refund, dispute policy (incl. USDC dispute process)
  • /mandates — Manage spending mandates (human-in-loop by default)
  • /embed — Shareable badges (Powered by MarketNow, Sentinel L2.5)
  • /pricing — Pricing tiers for sellers
  • /security — Sentinel L1.5 methodology
  • /handshake — Agent trust handshake protocol

Review Status Disclosure

Of 9,248 skills: 8,742 are auto-scanned (Sentinel ran, no human review), 22 are human-reviewed (curated free skills), 0 are maintainer-verified (program not yet launched). We never claim "verified" without earning it.

Maintained by AliceLabs LLC

Founded 2024 by Edison Flores (AliceLabs LLC, Wyoming USA). No investors, no employees besides the founder, no marketing budget. Built in public. Every change is a git commit visible at github.com/alicelabs-llc/marketnow.

Languages

English, Español, 中文, Português, Français

Keywords

MCP servers, Model Context Protocol, MCP marketplace, MCP directory, Claude Desktop skills, Cursor MCP, Cline MCP, agent skills, Sentinel security audit, Smithery, Anthropic MCP, open source MCP, MCP server registry, install MCP server