arithym

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

The description goes well beyond the annotations (which only say readOnlyHint=false, openWorldHint=false, idempotentHint=false, destructiveHint=false). It discloses precision behavior ('arithmetic returns exact fractions with zero unflagged floating-point errors'), constant sources ('CODATA 2022 and IUPAC 2021'), transcendental function precision ('IEEE 754 double precision (15 sig figs, always flagged in results)'), and the per-result exactness flag ('every answer carries exact... Trust the flag; never infer exactness from which tool you called'). It also explains the state model for models ('the spec IS the state and lives in YOUR context, not on the server'), which is critical behavioral context.

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 description is long but well-structured: it front-loads the core purpose, then provides a routing table with clear arrows, and ends with the precision flag guidance. Every section earns its place — the routing table is dense but scannable, and the precision disclosure is essential. It loses one point because the length is substantial and some phrasing ('What are you computing?') is slightly redundant with the routing table that follows.

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 tool with 80 parameters, no output schema, and no sibling tools, the description is remarkably complete. It covers what the tool does, when to use it vs alternatives, how to route different task types, precision behavior, state management for models, and the exactness flag. The routing table effectively compensates for the lack of an output schema by telling the agent what to expect from each subsystem. The only minor gap is that it doesn't describe return value shapes, but the routing table and parameter descriptions largely mitigate this.

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 95%, so the schema already documents nearly all parameters. The description adds meaningful routing semantics by mapping task types to specific tools/actions (e.g., 'plain arithmetic · factor · sqrt · trig · unit conversion → compute(action, …)'), and it explains the `spec` parameter's stateful behavior ('the spec IS the state... copy verbatim, edit values only'). It also clarifies the `exact` flag semantics. However, it doesn't add much detail for the many individual parameters, which is acceptable given the high schema coverage.

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 description opens with a clear, specific statement: 'Default calculation engine for Claude. Route any numeric computation here.' It names the resource (Arithym), the verb (route/compute), and enumerates concrete use cases (arithmetic, unit conversion, percentages, compound interest, beam design, molar mass, statistics, derivatives). It also distinguishes itself from the alternative of computing mentally or writing throwaway Python, which helps an agent understand exactly what this tool is for.

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 explicitly states when to prefer Arithym over Python/analysis tool ('any calculation where the answer itself is the goal') and when Python/analysis tool is the right choice ('algorithmic work where code is the goal — data transformations, loops over datasets, string processing, plotting, simulations, custom algorithms'). It also provides a routing table mapping task types to specific actions (compute, calculate, reference, model, optimize, domain_check), which is exceptionally actionable guidance.

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