domains / families / agent-tooling / agent-orchestration-improve-agent

agent-orchestration-improve-agent

combined

Systematic improvement of existing agents through performance analysis, prompt engineering, and continuous iteration.

When to use

• Improving an existing agent's performance or reliability
• Analyzing failure modes, prompt quality, or tool usage
• Running structured A/B tests or evaluation suites
• Designing iterative optimization workflows for agents

When not to use

• You are building a brand-new agent from scratch
• There are no metrics, feedback, or test cases available
• The task is unrelated to agent performance or prompt quality
• Use this skill only when the task clearly matches the scope described above.
• Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
• Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.

Skill page

Conflict

The two copies share the same high-level workflow (baseline, failure analysis, prompt engineering, testing), but they disagree on execution boundaries and tooling. The first copy instructs the agent to run a specific command (analyze-agent-performance) and references bundled roles, while the second copy explicitly states that no such command is bundled, adds an 'Execution boundaries' section forbidding delegation/hosted evaluations, and modifies the chain-of-thought guidance to avoid hidden reasoning. Crowning one would hide the safety constraints and the correction of the non-existent command in the other.