zabbix_hosts

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

Annotations already declare destructiveHint=true and readOnlyHint=false, so the safety profile is partly known; the description still adds material context they do not carry, namely that writes are two-step (preview first, then repeat with confirm_token) and that reads return compact rather than full records. This is exactly the kind of non-obvious behavioral detail an agent would otherwise get wrong on the first call. It falls short of 5 because it says nothing about required permissions, cascade behavior on deletes, or rate limits across the many methods.

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 method enumeration is dense and dominates the text, while the two highest-value sentences (compact reads, two-step writes, describe_method pointer) are buried at the end rather than front-loaded. Nothing is strictly filler, since each list maps to real callable methods, but an agent scanning the first sentence learns less than it should about the write-confirmation requirement.

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 three-parameter dispatcher with no output schema, an agent needs the method surface, the write protocol, and a path to per-method documentation; the description supplies all three. The remaining gap is return shape: 'reads return compact defaults' is asserted but never characterized, and with no output schema there is nothing else to fall back on.

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% (method, params, and confirm_token are all documented inline), so the schema already does the heavy lifting and a 3 is the baseline. The description reinforces confirm_token's role by tying it to the preview-then-confirm write flow, but it adds no new syntax, format, or example detail for the free-form 'params' object beyond the schema's own 'JSON-RPC params' note.

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 resource frame ('Zabbix hosts API') and then enumerates the concrete method families the tool can dispatch (host, hostgroup, hostinterface, template, templategroup, proxy, proxygroup, maintenance) with their allowed verbs. An agent can immediately tell this is a generic method-dispatch gateway covering host-adjacent objects, not a single-purpose operation. It loses a point because it never distinguishes itself from generic siblings such as zabbix_raw_api or zabbix_config, which plausibly do the same job.

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?

Usage is only implied by the enumerated methods: you would use it when you need a host/template/proxy/maintenance API call. There is no statement of when NOT to use it, and no routing to obviously overlapping siblings like search_hosts, get_host_details, list_host_groups, or manage_maintenance_window. The one explicit pointer, 'Use zabbix_describe_method for docs and parameters,' is a genuine routing rule but it addresses a prerequisite, not tool selection.

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