A tool description should answer five questions before an agent calls it
Summary
The author argues that a good tool description for AI agents should answer five key questions — when to use it, what each input means, what success looks like, all possible errors, and permission rules — since most tools only address one of the five, forcing the agent to guess the rest.
Similar Articles
Before I let an agent call a tool in prod, I ask these five things
A practitioner shares a pre-production checklist for letting AI agents call state-changing tools: clear principal identity, explicit delegation scope, tool listing disclosures, app readiness (idempotency, audit trails, per-agent auth), and fail-closed behavior when policy or auth is down.
Your agent doesn't need more tools — it needs a job description.
The author argues that AI agents don't need more tools—they need well-defined job descriptions: reusable roles bundling permissions and actions, plus encoded business logic so agents operate by company rules instead of improvising from prompts. Part of a series on making software agent-ready.
How to choose an AI gent tool? Sep 2026 edition
This article offers a practical framework for evaluating AI agent tools by focusing on key questions about job clarity, access needs, autonomy, maintenance, cost, and exit options, moving beyond feature comparisons.
How do you visualise agent tool use?
A discussion or tool for visualizing how AI agents use tools, likely to aid understanding and debugging of agent behavior.
How should agents handle those tools that are very useful for beginners but will limit their development in the later stages?
Discusses whether AI agents should recommend tools based on users' current needs or consider future scalability, and how to communicate potential long-term limitations.