ygo_ai_benchmark

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

With no annotations, the description carries the full burden and does a solid job: it discloses that first player alternates, that games hitting maxTurns are reported as unfinished and 'never as a win', what metrics come back (win rates, average game length, LP remaining), and that the AI is 'a policy, not a trained model — treat the number as a sanity check'. That last caveat is exactly the kind of expectation-setting that annotations would otherwise provide. It stops short of stating runtime/cost expectations or whether the run has any side effects on stored state.

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?

Four sentences, front-loaded with the action and scope, then the mode semantics, then the return values, then the interpretive caveat. Every sentence carries information an agent needs — nothing is repeated from the schema or 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 an 11-parameter tool with no annotations and no output schema, the description covers the essentials: what the run does, what the modes mean, and what comes back. The remaining gaps are the undocumented extraA/extraB arrays (presumably side decks) and no sense of how long a 50-game run takes, which matters given the tool's cost profile.

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 already 82%, so the baseline is 3, but the description genuinely adds meaning the schema does not: it explains that 'alternate'/'A'/'B' mean the AI faces 'the passive default responder' and that 'both' makes the result deck A's win rate vs deck B. The schema's own aiSide text only explains 'both'. It leaves extraA/extraB and exploration unaddressed, which are the main gaps.

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 specific verb and resource: 'Benchmark the heuristic duel AI over N seeded games on the real engine, alternating who goes first.' That is far more than a restatement of the name and conveys scope (seeded, N games, alternating first player). It does not, however, distinguish itself from adjacent duel-running siblings such as ygo_run_duel or ygo_simulate_duel_scenario, so an agent must infer the difference from the description alone.

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 concrete when-to-use guidance: aiSide 'both' is the mode to pick 'to test a custom card or deck change', and the other values are framed as AI-vs-passive-responder benchmarking. What is missing is an explicit exclusion — nothing says when to prefer ygo_simulate_duel_scenario or ygo_run_duel instead, so routing among the duel tools still requires inference.

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