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

With no annotations at all, the description carries the full burden for a generic proxy that can invoke arbitrary (potentially destructive) tools, yet says nothing about side effects, permissions, or reversibility. It also omits operational behavior an agent needs: what happens on an unknown tool name, whether arguments are validated against the target's schema, and whether the target tool's result is returned verbatim or wrapped.

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?

Two short sentences, zero padding, with the action stated first and the routing hint second. Every sentence earns its place.

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 unannotated, no-output-schema dispatcher, the description should cover argument conformance to the target schema, error semantics for bad names/arguments, and roughly what comes back. None of that is present, leaving real ambiguity for a tool whose behavior depends entirely on inputs.

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%, so 'name' and 'arguments' are already documented and the baseline of 3 applies. The description adds no meaning beyond the schema — notably it never states that 'arguments' must conform to the target tool's input schema, which is the single most important semantic constraint for this parameter.

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?

States a concrete verb+resource ('Call a tool by name with the given arguments') and implicitly distinguishes execution from discovery by pointing at search_tools. It never mentions get_tool_schema, the other obvious sibling, so an agent can't fully tell whether it needs to inspect a schema first.

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?

"Use this to execute tools discovered via search_tools" gives a clear triggering condition tied to a named sibling, which is more than most definitions offer. It gives no guidance on the prerequisite step of fetching argument shapes via get_tool_schema, and no exclusions or failure conditions.

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

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

The description is behaviorally rich — disk-first contract, default export directory, mandatory return of 'Saved to: <path>', parent-directory requirement — which goes well beyond the annotations. But it explicitly says the file 'is always written to disk', which conflicts with readOnlyHint=true; the description rescues the agent from being misled, yet the definition still contains an internal inconsistency an agent must resolve.

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?

Front-loaded with the purpose and the disk-first guarantee, and the recovery flow is scannable. It runs long for a four-parameter tool — the 'Similar to Google's Deep Research' sentence and part of the recovery flow restate usage that was already established.

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?

No output schema exists, and the description compensates fully by specifying the response shape (absolute path on the first line as 'Saved to: <path>') and the attached EmbeddedResource for GUI hosts. An agent knows what it will get back before calling.

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 100% schema coverage the baseline is 3, but the description adds real meaning: what each format actually produces (docx with headings/lists/TOC, markdown with citations, json with metadata) and the fallback behavior and location when output_path is omitted. This lets an agent choose a format without guessing.

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?

Specific verb set (save/export/download/archive), specific resource (a completed research session), and enumerated output formats. An agent can immediately distinguish it from research_deep or resume_research, which produce or resume the session rather than persist it.

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?

States exactly when to reach for it ('to recover your report after a research_deep run completes or is interrupted') and provides a numbered recovery flow showing the preceding sibling call and that no other arguments are required. Nothing about when-not or prerequisites is left to inference.

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

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

Goes beyond the annotations by disclosing the cost profile ('can incur two research workloads') and explicitly bounding behavior ('no synthesis, merging or scoring'), which matters given readOnlyHint=false and idempotentHint=false. Attribution of each report to its engine is also disclosed, though permissions and failure behavior when an engine errors are not covered.

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?

Two short sentences, both load-bearing: the first defines the operation and output, the second sets cost expectations and the division of responsibility. Nothing is redundant.

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 two-parameter, no-output-schema tool, the description covers operation, cost, and return contract ('attributed reports'). Missing only the error/degraded-result behavior when one of the two engines fails.

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 both parameters are already documented, including the 'exactly two different engine profiles' constraint in the engines description. The prose only reinforces the verbatim re-use of the query, adding little beyond the schema.

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?

States a specific verb (run the same question), a specific resource (two engines), and a concrete output (attributed reports). The 'exactly two' constraint distinguishes it from the single-engine research_run/research_deep siblings, though no sibling is named.

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 'exactly two engines' constraint implies the comparison use case, and the note that no synthesis is performed clarifies what the caller must do next. However, it never explicitly says when to choose this over research_run, research_deep, or research_web.

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

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

Annotations already declare readOnlyHint and openWorldHint, so safety is covered, and the description adds real value on top: a 3-20 minute latency envelope and automatic clarifying-question elicitation when the query is vague. It does not restate that mcp_servers is disabled or describe the collaborative_planning handoff, but those are carried by the schema descriptions.

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?

Front-loaded with purpose and latency, then the default/recommended use case, then the exclusion — good ordering for scanning. Minor waste in the model-version sentence ('fast/default April 2026 Deep Research agent') and stray line breaks, but nothing is genuinely misleading.

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?

An output schema exists so return values needn't be explained, and annotations cover the safety profile. Combined with the description's latency, use-case routing, sibling exclusion, and elicitation behavior, an agent has everything needed to decide to call it and what to expect while it runs.

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%, so the schema already documents every parameter including the disabled mcp_servers, the visualization enum, and the collaborative_planning follow-up call. The description adds only indirect param guidance (vague queries trigger elicitation) and no syntax or format detail, so the baseline 3 applies.

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?

States a specific verb+resource (autonomous research agent) with the scope-defining trait that matters most: 3-20 minute runtime. It explicitly names the sibling research_deep_max as the alternative for maximum comprehensiveness, so an agent can pick between the two without opening either 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?

Gives an explicit default rule ('use this by default for interactive questions, exploratory research, competitive analysis, comparisons, latency/cost-sensitive synthesis') plus an explicit when-not with the named substitute. It also discloses the fallback behavior for vague queries, which changes how an agent should phrase a call.

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

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

Annotations already declare readOnlyHint and openWorldHint, so safety is covered by structured data. The description adds beyond that: this is an autonomous agent trading speed for completeness, which implies long runtime and heavy web usage. It stops short of stating expected duration, cost, or that collaborative_planning returns a plan instead of a report (that detail lives only in the schema).

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?

Front-loaded with the core identity, then the when-to-use triggers, then the alternative. Efficient, though the trigger list is somewhat long and mixes user phrasing with internal scheduling terms ('offline/nightly') that could be tightened.

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?

An output schema exists, so return-value explanation is correctly omitted. Combined with a fully documented parameter schema and read-only annotations, the agent has nearly everything needed. The remaining gap is operational expectations: runtime, cost, and that this is a long-running autonomous job rather than a quick query.

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%, so each of the six parameters including the disabled mcp_servers and the collaborative_planning two-step flow is already documented in the schema. The description adds no additional parameter-level meaning, which is the baseline-3 case when the schema does the work.

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?

States a specific verb+resource+scope: an autonomous research agent run at maximum comprehensiveness. It immediately differentiates itself from the closest sibling by naming research_deep as the ordinary-depth alternative.

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?

Gives explicit trigger conditions (user says 'Max', 'deep research max', 'exhaustive', 'due diligence', 'literature review', 'board-ready', 'offline/nightly') plus the selection rule: maximum completeness over speed. It also names the exclusion case and the alternative tool for ordinary research.

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

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

Annotations already declare non-readOnly, non-idempotent, open-world behavior, so the bar is lower; the description still adds real value by disclosing that execution is asynchronous and that retries or resubmissions spawn new paid jobs. That cost/idempotency consequence is exactly the kind of context annotations cannot express. It stops short of describing latency, engine failure behavior, or what happens when a paid job errors.

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?

Short and front-loaded: the core action and its async nature come first, then the cost caveat, then the follow-up pointer. Three tight sentences with no filler, though the phrasing 'a retried or resubmitted Task is a new job' is slightly redundant with the preceding clause.

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?

With no output schema, the description correctly discloses that a run_id is returned and names the tool to consume it, which is the critical missing piece for a fire-and-forget job. Combined with the cost and async disclosures, an agent has enough to invoke it correctly; only error/failure semantics are absent.

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 both parameters (query, engine) are already fully documented with defaults and enum-like values in the schema. The description adds only the phrase 'selected engine' and confirms the query goes to a paid job — marginal value over the schema. Baseline 3 applies.

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?

States a specific verb and resource ('Run one research job on the selected engine') and adds the key trait that it executes as a background Task. It does not explicitly differentiate itself from fixed-engine siblings like research_deep or research_web, so the agent must infer that 'selected engine' is the differentiator.

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?

Gives a clear follow-up path ('Use get_research_run with the returned run_id to inspect or recover the result') and warns that each submission is a separately paid job, which implicitly discourages blind retries. However, it never states when to prefer research_deep/research_web/research_compare or when this generic engine-selecting entry point is the right choice.

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

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

Annotations already declare readOnlyHint and openWorldHint, so safety is covered. The description adds genuinely useful context beyond them: fixed high thinking level (cost/quality tradeoff) and latency varying with query and provider load. It stops short of saying anything about failure modes, cost, or rate limits for an open-world call.

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?

Front-loaded with the core behavior and return shape, then one line on quality/latency, then a trigger list. Efficient overall; the latency sentence is arguably secondary but earns its place as a cost signal.

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?

An output schema exists, so return-value detail is unnecessary, and annotations cover the safety/open-world profile. The description is complete enough to call correctly; only the relationship to the deep/compare siblings and any cost expectations are left unstated.

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% with only two parameters, both documented in the schema, so the baseline is 3. The description adds no meaning to either parameter — notably it says nothing about when a caller would want include_thoughts, nor how the query should be phrased (question vs keywords) for grounding.

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?

States a specific verb and resource ("Web research with Gemini grounding. Returns an answer with citations") and adds that it uses "a fixed high thinking level". It implicitly contrasts with the deep/deep_max siblings that presumably vary thinking level, but never names them, so sibling differentiation is only inferable.

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 "Use for:" list gives concrete, recognizable triggers — quick lookups, fact-checking, current events, docs, error messages, news. It lacks any when-not guidance or explicit pointer to research_deep / research_compare for heavier or comparative work, so the agent must guess at the escalation path.

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

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

Annotations already declare readOnlyHint and openWorldHint, but the description adds the crucial non-obvious behavior: the session is persisted at start, so research continues server-side after the client disconnects, and results are retrievable once ready. It also discloses the dual behavior of the two call modes. It stops short of saying what is returned while a session is still running, which is the one remaining behavioral gap.

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?

Front-loaded with the core action, then conditions, then the no-arg/id distinction, then the handoff. Every sentence carries information, though the paragraph is longer than strictly necessary for a single-optional-parameter tool.

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?

An output schema exists, so return values needn't be re-explained; the description instead covers the operational model that the schema cannot express. Both calling modes, the source of the id, and the downstream export path are all accounted for.

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 baseline is 3. The description goes beyond the schema by explaining where the interaction_id comes from (the original research_deep call or list_research_sessions) and what happens when it is omitted.

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?

States a specific verb and resource (resume a research_deep session) and scopes it to interrupted/in-progress sessions. It clearly distinguishes itself from research_deep (which starts a session) so an agent can route correctly without opening the 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?

Gives explicit trigger conditions (client disconnect, transport error, cancellation) and a second legitimate use case (polling whether a long-running session completed). It also spells out the no-argument vs interaction_id branches and names both the sibling that supplies the id (list_research_sessions) and the natural follow-up (export_research_session).

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

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

No annotations are provided, so the description carries the full disclosure burden. It usefully discloses the ranking behavior ('ranked by relevance') and the return shape, but says nothing about matching semantics (lexical vs semantic), result limits, or whether the search space covers all registered tools.

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?

Two short sentences with no filler, and the core action is front-loaded before the return-format note. Slightly more structure (e.g. a single sentence) would be equally clear.

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?

An output schema exists, so return values need not be explained, and the single parameter is fully documented. What remains missing is routing context: an agent cannot tell from this description when to search for a tool rather than directly invoke call_tool.

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% for the single 'query' parameter, so the schema already explains it. The description's 'natural language' phrasing adds only marginal semantics beyond the schema and gives no guidance on query formulation or specificity.

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?

States a specific verb and resource ('Search for tools') and adds the input modality ('natural language'), which is more than a restatement of the name. However, it never distinguishes itself from sibling call_tool, which is the obvious competing path for an agent that already knows which tool to invoke.

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 reference to 'the same format as list_tools' implies this is a filtered alternative to a full listing, but list_tools is not among the provided siblings, so that anchor may not resolve for the agent. No explicit when-to-use or when-not-to-use guidance is given relative to call_tool or the research_* tools.

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