# Tracert > The open route from agent intent to a provable outcome. Tracert is an open capability network: AI agents discover, evaluate and invoke free or pay-per-use services through one interface, with transparent evidence (TRACE Receipts) of what was promised, what was paid and what happened. TRACE = Transparent Registry for Agent Capabilities and Execution. Free capabilities never require a wallet or account. Quotes are exact and expire. Every execution reaches an explicit terminal state with machine-actionable reasons. Trust model: transparent evidence + buyer-side policy — no universal quality score. ## Machine entry points - [Registry index](https://tracert.site/index.json): every capability with status, pricing, media types and manifest URL - [TRACE Manifest schema](https://tracert.site/schemas/manifest/v0.1): JSON Schema 2020-12 for capability manifests - [TRACE Receipt schema](https://tracert.site/schemas/receipt/v0.1): JSON Schema for verifiable execution records - [Agent guide](https://tracert.site/agents): how to select, invoke and verify capabilities ## Capabilities - [ai-directory.publish-listing](https://tracert.site/capabilities/ai-directory.publish-listing): Submit an eligible AI tool website and receive a public directory listing page with a dofollow backlink, plus public repository evidence of the publication. (active, free; manifest: https://tracert.site/capabilities/ai-directory.publish-listing/manifest.json) ## For suppliers - [Publish a capability](https://tracert.site/publish): one bounded promise per TRACE Manifest, submitted by pull request to the public registry at https://github.com/tracert-network/tracert; free listings are first-class ## For agent developers - [Connect your agent](https://tracert.site/use): the five-tool MCP router (search_capabilities, get_capability, get_quote, invoke_capability, get_execution); local install today, hosted endpoint coming - [About the network](https://tracert.site/about): lifecycle, trust model, standards posture (MCP, A2A, MPP, x402, AP2)