diagnose_job

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations cover the safety profile (readOnly, idempotent, non-destructive), and the description adds genuinely non-obvious behavior: findings are ranked, and Prometheus-derived trends are explicitly 'observations with their metric and window, not causes.' That last clause prevents an agent from misattributing a root cause, which is behavioral context unavailable anywhere else.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the verdict-and-when-to-use message before the conditional Prometheus detail. Three tight sentences, each introducing distinct information (what it combines, when to call it, optional trend behavior and its epistemic caveat) with nothing redundant.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

An output schema exists, so return values needn't be explained, and the description covers the main diagnostic scope plus the optional history mode well. The only gap is the undocumented `cluster` parameter — relevant for a multi-cluster toolset alongside list_clusters and fleet_overview.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 0%, so the description carries the burden. It fully explains `history` (default '1h', 'off' to skip, and what it reports: restarts, checkpoint failures/durations, sustained backpressure), which is the least obvious parameter. But `cluster` and `job_id` get no mention at all, leaving one third of the surface undocumented.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific resource ('a job') and action ('one-call health verdict'), and enumerates the inputs it synthesizes: state, task status, exceptions, checkpoints, backpressure, source lag. This clearly distinguishes it from narrow siblings like job_exceptions or checkpoint_stats, which surface only one of those signals.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicitly routes the agent: 'Start here when asked "why is this job unhealthy/slow?"' — a concrete triggering condition, with the alternative interpretation (drill into single-signal sibling tools) implied by the enumerated signals. It also states the exclusion for Prometheus: a Prometheus problem never changes the live diagnosis.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.