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

Rich behavioral disclosure well beyond the annotations: it states the exact output categories (STRONG VALUE / FAIR / WEAK VALUE / INSUFFICIENT DATA), the itemized-and-scored evidence format, graceful degradation ("A source that fails is noted, not fatal"), data limitations ("Scorecard earnings cover federally-aided students only and lag by years"), and the advisory boundary ("not admissions, financial, or career advice"). No contradiction with readOnlyHint/idempotentHint — the description explicitly calls it a "plain read."

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?

The ~110-word description is front-loaded with the core value proposition, then builds logically: data sources, output format, failure behavior, caveats, disclaimer. Every sentence earns its place — the parenthetical data-source details and caveats are substantive for a synthesis tool, not padding. Nothing is redundant with the schema.

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?

For a complex cross-source synthesis tool with no output schema, the description carries the return-value burden and does so thoroughly: it names the verdict categories, explains what evidence is itemized, discloses scope limits (US, federally-aided students), and covers failure behavior. An agent can predict both the call's inputs and its response shape without opening any further documentation.

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 description coverage is 100%, so the schema already documents all three parameters well, including program's role in adding "program-level debt-vs-earnings evidence" and state's disambiguation purpose. The description adds only marginal specificity — that program-level evidence consists of "median debt and 1-year earnings" — but mostly restates what the schema already conveys. Baseline 3 is appropriate; the description does not meaningfully compensate beyond the schema.

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?

The opening phrase "One-call 'is this degree worth the cost' read for a US college (and optionally a named program)" states a specific verb (read/evaluate) with a specific resource (US college + program) and a crisp value proposition. It clearly distinguishes itself from siblings like college_metrics (raw metrics), college_compare (side-by-side comparison), college_outcomes_by_program (raw program data), and college_search (lookup), by emphasizing it is the synthesis/verdict tool rather than a raw-data tool.

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?

The description implies usage context well — this is the one-call value judgment tool, and the phrase "when a program is named" explains the optional-program behavior. However, it never names sibling alternatives or states when to prefer college_metrics, college_compare, or college_outcomes_by_program instead, so an agent must infer the routing decision from purpose rather than receiving explicit when-to-use/when-not-to-use guidance. The "informational only" disclaimer is the only explicit exclusion.

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