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

Annotations already declare destructiveHint=true and idempotentHint=false, but the description adds real behavioral value by disclosing the confirm=false preview / confirm=true apply mechanism — critical for a destructive write. It also explains the downstream effect on Smart Bidding. It stops short of documenting permissions/auth requirements or repeat-adjustment behavior.

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 core purpose in the first clause, then organized into bullets and a field list that are easy to scan. It is longer than average but each block earns its place (mode semantics, input format, field semantics, preview flag); the value=1200 vs value in example formatting is slightly redundant but minor.

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 need no explanation, and the description covers the hard parts of a complex write tool: operation modes, free-text encoding format, required versus conditional fields, and the dry-run gate. Only the customer_id and currency_code parameters are left entirely to an undocumented schema.

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?

With 0% schema coverage, the description carries the load and largely does so: it defines conversion_action (name or ID), the two accepted formats for the opaque 'adjustments' string, the full field list with which are required (datetime, conversion_datetime needed with gclid), identification strategy (order_id recommended) and worked examples. customer_id and currency_code remain undocumented, so it is not fully compensating for the coverage gap.

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 ('ADJUSTS conversions you already reported — the refund/cancellation half of the closed loop') and immediately disambiguates the two operations RETRACT vs RESTATE. This distinguishes it from the sibling upload tools (upload_offline_conversions, upload_microsoft_offline_conversions) that report new conversions rather than correcting existing ones.

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 scopes usage to conversions already reported and gives concrete when-to-use for each mode (RETRACT for refund/cancelled order/unqualified lead, RESTATE for partial refund/upsell), plus the business reason ('otherwise Smart Bidding keeps optimizing toward revenue that never landed'). It does not explicitly name the sibling alternative for net-new uploads, so the exclusion is inferable rather than stated.

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