AI Agents
What are Prodgy Agents?
Section titled “What are Prodgy Agents?”Prodgy agents are specialized tools that execute specific processes using AI. Each agent is configured within the platform with its own system prompt, LLM provider, tools, and integrations — designed to execute tasks such as generating test plans, analyzing code, creating documentation, or prioritizing backlogs.
Discovering Agents
Section titled “Discovering Agents”Use list_agents to see all available agents for your product:
"List available Prodgy agents"
The response includes each agent’s name, description, and ID — so your AI assistant can understand what each agent does and choose the right one.
Executing Agents
Section titled “Executing Agents”Use invoke_agent to delegate a task to a Prodgy agent:
| Parameter | Required | Description |
|---|---|---|
query | Yes | The task or question to send to the agent |
agent_id | No | Specific agent ID (from list_agents). If omitted, Prodgy auto-selects the best agent |
Examples
Section titled “Examples”"Run the backlog prioritization agent with: Prioritize the current sprint backlog based on business value and dependencies"
"Invoke the test plan agent: Generate a test plan for the new authentication module"
When to use agents vs. doing it yourself
Section titled “When to use agents vs. doing it yourself”| Scenario | Approach |
|---|---|
| Task requires your project’s specific AI process | Use invoke_agent — the agent has its own LLM, system prompt, and tools configured in Prodgy |
| Task requires access to external services via integrations | Use invoke_agent — agents can call integrations configured in Prodgy |
| Simple code generation or analysis | Do it yourself — your coding assistant is already great at this |
| Task requires knowledge from the knowledge base | Either — you can search yourself with search_knowledge or let an agent do it |
Auto-selection
Section titled “Auto-selection”If you don’t specify an agent_id, Prodgy will automatically select the most appropriate agent based on your query. This works well for general tasks, but for recurring or specialized tasks, specifying the agent ID ensures consistency.