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

Annotations only say readOnlyHint=false, openWorldHint=false, destructiveHint=false; the description carries the real behavioral burden and does so richly: rejection with supported-alternative naming, version pinning/substitution, installation on the project VM, persistence of resolved versions, and the slow-then-fast two-call persistence pattern. No contradiction with annotations — mutation (readOnly=false) matches 'Add npm packages', and nothing suggests destructiveness.

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 dense (roughly a hundred words across five sentences), but every sentence earns its place given the 0% schema coverage: purpose, exclusions, version pinning, install target, and the two-call pattern. It is front-loaded with the core purpose. It sits at the upper edge of appropriate length, with the kysely detail being specific but justifiably high-value.

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 two-parameter mutating tool with no output schema, the description covers validation, rejection behavior, exclusion rules, version pinning, install location, persistence, and the async job-then-recall flow. Minor gaps remain: it does not explain what the response/return looks like, how to correlate with job status, or the rationale for the second call beyond persistence, which would round it out.

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 0%, so the description must compensate — and it largely does. The `packages` parameter semantics are substantially enriched: bare names install pinned versions, explicit name@version overrides are for deliberate upgrades, and packages are validated against a supported set. `projectId` is only implied as the target project/VM ('Installs on the project VM'), leaving its meaning to inference rather than explicit statement.

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 specific verb and resource: 'Add npm packages to the project', and immediately adds scope ('validated against Floot's supported set'). It is clearly distinguishable from the sibling remove_dependency and from file-edit tools, and the 'add_dependency again' self-reference reinforces its exact role. No tautology or vagueness.

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?

Strong when/when-not guidance is present: avoid node-gyp/native packages, WASM modules, and large-binary bundles, with explicit exceptions and a version-nuance example for kysely. It also gives an invocation pattern for the async two-call flow. However, it never names alternatives (e.g., remove_dependency for removal, or the job-status tool for monitoring the slow install), so the 'vs alternatives' part is implicit rather than explicit.

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