Detect Zero-Iteration Failure
This skill provides a workflow for identifying and classifying agent failures that occur prior to any task execution.
Identification Criteria
Flag a task execution as a Zero-Iteration Failure if all of the following conditions are met:
- Iteration Count: The agent reports 0 completed iterations.
- Tool Usage: No tools were invoked during the session.
- Artifacts: No documents, files, or code artifacts were created.
- Conversation Log: The log ends abruptly or contains only an initial error message without agent reasoning steps.
Diagnosis Strategy
When a Zero-Iteration Failure is detected, do not attempt standard agent-level debugging (e.g., analyzing reasoning chains or tool parameters). Instead, investigate system-level causes:
- Environment Health: Check if the runtime environment initialized correctly (e.g., dependencies installed, API keys loaded).
- Prompt Parsing: Verify if the initial user instruction was malformed or triggered safety filters immediately.
- Initialization Crash: Look for stack traces or panic logs occurring before the first agent step.
- Resource Limits: Check if memory or quota limits were hit before execution began.
Action Plan
- Tag the Failure: Label the incident as
failure_mode: pre_executionrather thanfailure_mode: agent_logic. - Escalate: Route to infrastructure or platform engineering rather than prompt engineering.
- Retry Strategy: If automating retries, reset the environment context before retrying, as the previous context may be poisoned.
Example Indicators
textStatus: Failed Iterations: 0 Tools Called: [] Error: "Session initialization failed: Missing environment variable"
