Developer Workflow

Step 1: Setting Up the Development Environment

Developers begin by accessing the Agentify Developer Framework (ADF):

• Download the ADF via CLI, SDK (JavaScript/Python), or a web-based IDE.

• Review documentation and tutorials, including boilerplate templates (e.g., a basic Swap Agent or a monitoring template).

• Set up dependencies like Node.js for JavaScript SDK or relevant Python libraries.

This step ensures developers have a ready-to-code and test environment tailored for agent creation.

---

Step 2: Designing the Agent

After setup, developers design their agent:

• Purpose: Define what the agent does, e.g., a "Rebalance Agent" that maintains a user’s asset ratio.

• Structure:

  • Triggers: User commands or time-based automation.

  • Actions: Use internal libraries or call other agents like the Swap Agent.

  • Logic: Workflow for analyzing data and executing multi-step operations.

Developers can implement agents via code or prompt-based logic (for LLM guidance).

---

Step 3: Testing in the Agentify Sandbox

Once logic is implemented, testing is done using the Agentify sandbox:

• Simulate inputs (e.g., portfolio rebalancing scenarios).

• Observe dry-run outputs like proposed swaps or rebalances.

• Test edge cases: balanced portfolios, insufficient funds, etc.

The sandbox provides verbose logs and debugging insights including:

• Parsed intents

• Chosen execution paths

• Error messages for invalid logic

---

Step 4: Security and Compliance Check

Before deployment, developers perform security audits:

• Ensure safe protocol calls (via whitelisted APIs or verified contract addresses).

• Avoid hardcoded secrets or brittle assumptions.

• If supported, run unit/integration tests using the ADF test suite.

• Validate any external APIs (e.g., price feeds) for security and consistency.

---

Step 5: Deployment to the Agentify Marketplace

With the agent finalized:

• Developers package and deploy using ADF tools.

• Required metadata:

  • Agent name, description, icon, tags, category

  • Pricing model (e.g., free, fee per call in $AGNT)

  • Optional developer stake (as bond) in $AGNT

The agent is registered via a smart contract. Key data includes:

• Agent logic (or IPFS hash)

• Developer address

• Monetization rules

Agents undergo automatic or manual review for malicious patterns before listing.

---

Step 6: Publication and Verification

Once approved:

• The agent appears on the Agentify Marketplace (with a "New" label).

• Developers receive an Agent ID and sharable URL.

• Agents can be tested live by the developer on testnets or with low-value mainnet funds.

User feedback and activity stats are monitored through the dashboard. Developers may:

• Request audits or community verification

• Gain trust via badges and reviews

---

Step 7: Maintenance and Updates

Ongoing improvements and fixes are streamlined:

• Developers push new versions with the ADF deployment tool.

• Users are notified of updates, or agents auto-update depending on the security model.

In case of bugs:

• Quick patches are pushed

• Agents can be temporarily disabled by developers or by Agentify Core if necessary