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

Annotations already flag destructiveHint=true and idempotentHint=true, but the description adds valuable behavioral detail: it returns only the resulting unread count, not per-id results, and explains the security rationale (avoiding an enumeration oracle). It reinforces destructiveness with 'This cannot be undone.' This goes beyond the annotations and explains the response format and why, which is exactly the kind of transparency needed. It doesn't cover idempotency explicitly, but that's already in the annotations, so no 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 description is front-loaded with purpose in the first sentence, then covers irreversibility, then explains the return behavior and rationale. It is about 40 words, and every sentence serves a clear purpose: the return behavior explanation is important for the agent to set expectations. There is no fluff or tautology, though it could be slightly more compact, but the detail is justified.

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 one-parameter destructive tool, the description is quite complete: it states what it does, what it returns (unread count), and that it's irreversible. The schema covers the ignore behavior for invalid ids. It doesn't explicitly mention idempotency or auth, but those are either covered by annotations (idempotentHint) or standard for the platform. The only gap is explicit guidance on when to use this vs. alternatives, which is more a usage-guidelines concern. Overall, it's complete enough for an agent to call it correctly.

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% (the ids parameter has a descriptive text). The tool description adds only that it is 'a chosen set,' which is redundant with 'ids' and 'delete a chosen set.' Since the schema already explains the parameter well (max 100, ignoring nonexistent or others' ids), the description adds minimal semantic value. Baseline 3 is appropriate given full 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 states exactly what the tool does: 'Delete a chosen set of your notifications.' This is specific about the verb (delete), resource (notifications), and scope (a chosen set, implying user-provided ids). It also adds that it cannot be undone, sharply distinguishing it from read-marking operations. The 'chosen set' phrasing clearly sets it apart from siblings like colony_delete_read_notifications (which deletes all read) or colony_delete_notification (singular), making its purpose unambiguous.

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 when you have a specific set of notification ids to delete, saying 'a chosen set of your notifications.' However, it does not explicitly name alternatives like colony_delete_read_notifications or colony_delete_notification, nor does it state when not to use this tool. The guidance is implied rather than explicit, so an agent must infer from context which sibling to choose, though the phrase 'chosen set' is a reasonable hint.

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