TopazTOPAZDocs

Agents

Working with Topaz from AI agents

Topaz publishes an official agent skill that lets any AI agent quote, build, and reason about swaps, pools, gauges, bribes, veTOPAZ, and rewards on BNB Chain. This section covers how to load it, what it enables, and how the safety model works.

What the skill is

The Topaz agent skill is a runtime-neutral package — markdown instructions, protocol references, worked examples, evals, and TypeScript helpers — that gives an AI agent the structured context it needs to act safely on the Topaz protocol. It is not a chatbot, API, or hosted service. It is a set of files that any agent can read and follow.

The canonical source lives at github.com/topazdex/agent-skill. The website mirrors /skill.md and /skill.json dynamically from that repo, so an agent can load the latest version from either source.

ℹIf you're shipping an agent
The fastest place to start is the /agents page — quick start commands, entry-point URLs, and the installer in one place. The docs section you're in now provides longer-form guides for specific topics.

Who this is for

  • ✓Developers shipping AI agents that need to interact with Topaz — trading bots, dashboard copilots, governance helpers, on-chain analytics agents.
  • ✓Power users of generalist agents (Claude Code, Codex, Cursor, Hermes, OpenCode) who want their agent to understand Topaz workflows natively.
  • ✓Integrators building agent-facing tooling — quote widgets, swap calldata generators, voting assistants — and want the protocol model built in.

Three ways to load the skill

The skill is designed to plug into whatever loading mechanism your agent supports. Pick the one that matches.

  • ✓Read from URL — best for browser agents, chat agents, or anything with a web-fetch tool. Point at topazdex.com/skill.md and you're done.
  • ✓Clone the full package — best for coding agents and CLI agents. Includes references, examples, scripts, and TypeScript helpers.
  • ✓Copy into a native skill system — for agents with their own skill, rule, or memory concept, drop the package in and point at the repo for support files.

The Runtimes & Install page walks through each of these per runtime with the exact commands and prompt templates.

What the skill enables

Every protocol workflow on the Topaz app is also reachable via the agent skill: swaps and quotes, v2 LP and v3 NFT positions, gauges and emissions, veTOPAZ locks and voting, bribes and reward claims, plus subgraph and on-chain reads. See the Capabilities page for example prompts grouped by workflow.

Safety, by default

The skill is designed so agents quote, explain, build, and simulate by default — and only broadcast transactions when the user explicitly asks and signing credentials are configured. Every output an agent produces is labeled (quote, built calldata, approval-needed, or broadcast tx-hash) so the user always knows whether anything actually moved.

The full safety model — including the four output labels and when each is appropriate — is on the Safety Model page.

Continue reading

Runtimes & Install →

Per-runtime install paths, prompts, and the auto-detect installer.

Capabilities →

Example prompts grouped by workflow — quote, swap, vote, claim, lock.

Safety Model →

The four output labels, when broadcasting is OK, and how to enforce read-only.

/agents (entry page) →

Quick start commands, entry-point URLs, and the one-line installer.