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

The description goes far beyond the annotations by disclosing the two-phase execution (build-then-confirm), the verification eagerly awaited once verify=true, the "verification" block semantics ({confirmed, slot, post_state, expected_vs_actual}), and the explicit instruction to gate on verification.confirmed rather than tx_signature. It also reveals fee billing details (x402 payment_headers, pay_for_the_block discount) and confirms the non-custodial flow. This substantially adds behavioral context that annotations alone (only read-only hints) already communicate.

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 organized in cohesive, rule-structured blocks: purpose+payload, fees, workflow routing, idempotency, and verification protocol. Every section earns its place since the two-phase mechanism and idempotency dedup are genuinely non-obvious, and the text keeps each section compact. It pulls the main action-parameter details first and backloads the complex semantics, which is ideal for an agent sampling the description.

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 10-parameter, mutation-safe tool with a rich output schema, the description is impressively complete: required parameters, units, protocol values, fee behavior, static text and verification timing are all covered. The remaining gap is `allow_unverified` (and to a lesser extent `caller_id`), which is left to the schema's default. Since the description already carries the most safety-critical instructions, one missing parameter is a minor but real completeness shortfall.

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 schema description coverage at 0%, the description is the primary source of parameter meaning and it carries that burden well: it documents amounts=(lamports), protocol=(enumeration), verify=skipping the confirmation wait, idempotency_key (client UUID, dedup and replay semantics, same-key-across-legs rule), signed_transaction (the second-phase re-call), and the pay_in_crank/payment_header pairing. The only noticeable gap is `allow_unverified`, which is left undefined; its behavior appears important for results variation but is undeclared.

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 + resource + outcome: "Stake SOL for a liquid-staking token (non-custodial)." It then names the distinct protocols (marinade | jito | blaze), clarifies the return payload (unsigned base64 tx, LST, APY), and clearly separates this stake action from other operations. An agent can tell exactly what to expect from this tool at a glance.

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 explicitly positions the tool in a workflow: "EXECUTE step (yield leg)" and instructs to "stake the idle slice after comparing the yield rates get_lst_yields", plus "Monitor via portfolio_snapshot. See get_trading_workflow." This gives the AI encoded context on when — and in what order — to invoke it. It does not list explicit exclusion warnings, e.g., when to choose an alternative like lst_swap, so it misses a bit of against-criteria guidance.

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