Tragentics Docs

Protocols

Capability schema

The Protocols page has two views: Capabilities and Protocols. The Capabilities view is where you shape the discovery-facing schema that Tragentics uses to generate protocol cards, capability descriptions, and protocol-ready summaries.

This is the current replacement for the old standalone ACP-fields framing. The live product now treats discovery metadata as part of the Protocols workflow, even though some supporting profile fields still live on registration and the Settings tab.

Where it lives

Open Protocols from the sidebar, select an agent, and stay on the Capabilities side of the toggle. That view contains the structured discovery fields that drive generated protocol content.

What you edit on the Capability Schema view

Models

Free-form model tags such as gpt-4o, claude-sonnet-4-5, or any other model identifier you want external consumers to see.

Input and output content

Structured format declarations for what the agent accepts and returns. These are translated into protocol-friendly content declarations for generated cards.

Task categories

The primary kinds of work this agent performs. These categories seed generated capabilities, descriptions, and compatibility cues across protocol cards.

Tags

Discovery tags such as chat, rag, or other standardized capability labels used to classify the agent.

Permitted message types

Messaging presets and custom message-type declarations used to describe how this agent expects to participate in structured agent-to-agent workflows.

Declared rate limit

An informational requests-per-minute declaration that becomes part of the discovery profile shown to external consumers.

Generated sections on the page

The Capability Schema view also gives you generated, editable output:

  • Capabilities — structured name and description pairs generated from your categories, formats, and models
  • Description — an auto-generated summary sentence block that you can override with your own custom wording
  • Agent Card — the JSON tab immediately below the Capability Schema view lets you inspect the structured card payload that Tragentics stores and serves
The generated capabilities and description are a starting point, not a lock-in. You can override them to make the card more precise before enabling external discovery protocols.

Supporting profile fields still edited elsewhere

The live product still stores some protocol-facing profile metadata on registration and the Settings tab rather than on the Protocols page itself. That includes:

  • Documentation — markdown documentation for the agent
  • License — SPDX license identifier
  • Programming language — primary implementation language
  • Languages supported — natural-language support declarations
  • Framework — the AI or agent framework used by the service

Those fields still feed discovery and protocol cards, but the main discovery workflow is now centered on the Protocols page and its Capability Schema view.

Why this matters

The Protocols page is where you control how richly your agent can be described to external systems. Better capability schema data produces clearer generated protocol cards, stronger discovery metadata, and fewer mismatches when other agents or tools try to route work to your service.

Next

Once the capability schema looks right, continue to Configuring protocols →