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

Annotations already cover safety (readOnlyHint, idempotentHint, destructiveHint=false), and the description adds substantial non-obvious behavior on top: the served projection is dynamic ('Read `_meta.available_edge_types` — computed from the served projection on every call — before assuming an edge type exists'), the edge-type list is a 'HINT, not the authority,' a missing node type is 'a documented boundary and never a silent gap,' and `direction` is ignored without `node_id` and 'the response says so rather than pretending it applied.' The historical note about the stale four-name list concretely justifies why the agent must trust the meta-field over the static docs. No contradiction with annotations.

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 long, but it is front-loaded with purpose and every sentence carries a distinct piece of information — purpose, exclusivity, scope, boundary behavior, invocation modes, narrowing parameters, and the meta-field warning. The only mild excess is the historical anecdote about the four-name list, but it earns its place by justifying the 'hint, not authority' instruction. For a tool with 7 parameters, two invocation modes, and a complex server context, the density is appropriate rather than padded.

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 tool of this complexity — dual node/edge modes, cursor pagination, dynamic meta-fields, and a server where it is the sole edge-serving tool — nothing an agent needs to call it correctly is missing. The tricky parts are all addressed: the boundary of the served projection, how to confirm available edge/node types, the node_id format, and the mode-switching parameter interplay. An output schema exists, so return-value details need not be in the description.

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 coverage is 100%, so the schema already documents every parameter in detail — baseline is 3. The description adds genuine value above that: concrete node_id formats (`module:M178`, `concept:C001`, `study:ai-impact-2026-EN`, `vendor:alteryx`), the mode-switching semantics of `node_id` (edges vs nodes), and the find-id-first workflow linking `node_type`/`query` to `node_id`. It stops short of restating what the rich schema descriptions already say, which is exactly the right division of labor.

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 names a specific verb and resource — query the relations/edges of the learning knowledge graph — and immediately distinguishes it from everything else on the server: 'THIS IS THE ONLY TOOL ON THIS SERVER THAT SERVES EDGES; the others serve rows.' It further refines the scope to four carried node types, so an agent knows exactly what this tool is and is not. This is exemplary sibling differentiation without opening any schema.

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?

It gives an explicit when-to-use rule ('Ask it what connects to what, not what exists'), names the contrast with the row-serving tools, and explains when NOT to use it: the seven unserved node types are 'each either served by their own tool or named as not served at all,' with `_meta.excluded_node_types` as the authority. It also lays out the two invocation modes — `node_id` to walk a neighbourhood, `node_type`/`query` to find a node id first — which is concrete, actionable routing guidance.

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