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

With readOnlyHint=false already set, the description still adds substantial disclosure: sign-in requirement, that a new bill replaces the earlier one for the same service/month (idempotent, matching idempotentHint), that remove deletes a month or a single service, that alerts can trigger an outbound email, and that a >20%/$5 gap gets flagged. Note the mild tension with destructiveHint=false given 'remove deletes a month', but this is record-level cleanup of bill entries rather than account data loss, so it does not read as a contradiction.

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 content is dense and mostly earns its place, but it arrives as one long run-on paragraph with stacked parentheticals, so key facts (mode selection, replacement semantics, return shape) are hard to scan. The trailing 'How to present' instruction is useful but adds further bulk without any structural separation.

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 7-parameter, nested-object, multi-modal tool with an output schema, this covers auth, write semantics, side effects and mode selection thoroughly. Return values are re-described despite an output schema existing, which is redundant rather than harmful; the remaining gap is error/partial-failure behaviour for CSV or invoice parsing.

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?

Goes well past the 71% schema coverage: it specifies the bills item shape (month as YYYY-MM, vendor_id or invoice name, amount_usd), what columns the csv must contain, that invoice_text is pasted PDF/email text from which service/month/total are extracted, and that preview only reads without saving. It also clarifies the id-absent behaviour and the month default, which the schema only partially carries.

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 verb and resource ('Record what the user actually paid each service in a month, on one of their saved stacks or audits') and the downstream comparison against the estimate. An agent can distinguish this from estimate_cost, plan_stack or project_stack without opening the schema, since this is the only one that ingests actual invoices/bills.

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?

Clearly routes the three input modes (bills, csv, invoice_text) and the id-less variant ('the bills are kept on the account, not on a stack'), plus the preview/remove sub-modes. It also tells the agent to offer update_saved_stack afterwards when the real user count differs. It never explicitly says when NOT to use it (e.g. use estimate_cost for hypotheticals), but the context is unambiguous.

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