kit

AI Agent Starter Kit — Agent-Ops Harness

Python agent-ops harness for Claude: BaseAgent, Redis memory, a proposal engine, and Ops / Content / Support agents. Examples: single_agent.py and multi_agent.py. No LangChain, CrewAI, Docker, or 12-tool pack in the zip.

What's included

✓BaseAgent with Anthropic API, exponential-backoff retries, and Pydantic tool schemas
✓Three starter agents: Ops (health checks), Content (blog/social drafts), Support (inbox drafts)
✓Proposal engine — agents propose actions; humans approve before execution
✓Company state in Redis: shared namespaces, task queues, and event streams
✓Redis-backed conversation memory, audit log, and per-cycle performance scores
✓Slack webhook reporting when slack_webhook_url / SLACK_WEBHOOK_URL is set
✓Examples: examples/single_agent.py and examples/multi_agent.py — Redis via config redis_url

Code sample

python

from agents.base_agent import AgentTool, BaseAgent

class ResearchAgent(BaseAgent):
    def get_system_prompt(self) -> str:
        return "You are a research assistant."

    def get_tools(self) -> list[AgentTool]:
        return [
            AgentTool(
                name="take_note",
                description="Record a research note.",
                input_schema={
                    "type": "object",
                    "properties": {
                        "subtopic": {"type": "string"},
                        "content": {"type": "string"},
                    },
                    "required": ["subtopic", "content"],
                },
                handler=self._take_note,
            ),
        ]

    async def _take_note(self, subtopic: str, content: str) -> dict:
        return {"stored": True, "subtopic": subtopic}

# Requires Redis + ANTHROPIC_API_KEY (see config.example.yaml)
# agent = ResearchAgent(name="research-agent", config_path="config.yaml")
# await agent.initialize()
# await agent.run_cycle(task="Research AI automation")

$47

Instant download · Lifetime access

Get it — $47

30-day money-back guarantee

kit