Skip to content

examples.md

Latest commit

 

History

History
43 lines (39 loc) · 2.19 KB

examples.md

File metadata and controls

43 lines (39 loc) · 2.19 KB
Raw

Examples

The examples/ directory in the Blocklog Python SDK contains runnable scripts demonstrating the core usage patterns. They reflect the current public APIs and are actively tested.

01_quickstart.py

Purpose: Demonstrates the bare minimum to trace an agent and record a decision. Features demonstrated:

  • SDK Initialization (blocklog.init())
  • Decorating a tool (@blocklog.tool)
  • Decorating an agent (@blocklog.agent)
  • Recording a decision (blocklog.decision()) Expected Output: Logs demonstrating the initialization of the agent. The output will conclude with a success message containing the decision UUID generated:
Decision recorded: <uuid>
Done. Check your Blocklog dashboard.

02_stock_trading_agent.py

Purpose: Simulates a fully functional trading agent that fetches market data, evaluates momentum, and conditionally executes a trade. Features demonstrated:

  • Advanced tool metadata and tagging (tags=["market-data"])
  • Contextual tracking of outputs: Requesting human approval for trades exceeding $8k threshold via d.request_approval()
  • Immediate cryptographic verification of the recorded decision via blocklog.verify.decision(decision_id) Expected Output: A detailed execution trace printed to the terminal of the stock-trader agent checking the market price and placing an order. A decision record is produced and verified:
  ✓ Decision recorded: <uuid>
  ✓ Verification status: verified

03_multi_agent_workflow.py

Purpose: Demonstrates a complex, multi-agent pipeline sharing a common workflow_id. Features demonstrated:

  • Multi-agent correlation (Analyst → Risk Manager → Execution Engine)
  • Passing context dynamically using metadata={"workflow_id": WORKFLOW_ID}
  • Generating a compliance report for the entire workflow via blocklog.compliance.generate() Expected Output: Three distinct agent executions logging their steps to the console. A chain of decisions is created sequentially (SIGNAL → RISK_APPROVAL → TRADE_EXECUTION). Finally, the terminal outputs the report_id of the SOC2 compliance report:
[Orchestrator] Generating compliance report...
[Orchestrator] Compliance report: <report_id>