Error-actionability translation for MCP tool failures

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

With no annotations, the description carries the full burden, but it does disclose the payload contents (counts and hit/miss totals), implying a read-only aggregate query. It says nothing about scoping to 'this account', freshness, caching, or whether data spans all servers by default.

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?

A single compact phrase with no filler, front-loading the scope ('this account's') and the resource. It is a fragment rather than a sentence, which slightly weakens clarity.

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 no-parameter, no-output-schema tool, describing the returned metrics is the right move and largely covers the gap. Missing are the response structure (per-server keying, field names) and any access or freshness caveats an agent would need to trust the numbers.

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?

The tool takes zero parameters, so there is nothing for the description to clarify beyond the schema. Baseline 4 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?

The description identifies a specific resource – per-server dialect profile counts plus hit/miss totals – which is concrete enough for an agent to know what data comes back. It lacks an action verb and never names how it differs from siblings like verdict_stats, so it falls short of a 5.

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?

There is no indication of when to call this versus verdict_stats or the error_* tools, and no exclusions or prerequisites. The agent must infer usage entirely from the resource name.

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 must carry the full behavioral burden. It hints that the tool 'learns the per-server dialect profile,' suggesting stateful behavior, but does not explain whether this writes data, requires prior profiles, or has side effects.

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 a single compact sentence with the core purpose front-loaded. It is dense and somewhat jargon-heavy, but it does not waste space.

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?

The tool has five parameters, four of which are required, a nested object, no annotations, and no output schema. The description gives only a high-level purpose and omits enough parameter, return-format, and side-effect detail to make invocation reliably clear.

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 0% across five parameters, and the description only loosely maps to 'prose' and 'server' while leaving 'tool,' 'mutating,' and the nested 'structured' object unexplained. It does not compensate for the schema's lack of parameter documentation.

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 states a specific verb and resource: it classifies one MCP tool-failure prose into a recovery verdict. It also mentions the per-server dialect profile, but does not explicitly differentiate itself from siblings such as error_enforce or dialect_profiles.

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 phrase 'one MCP tool-failure prose' implies that this is used for individual failure classification, but there is no explicit guidance on when to use this tool versus error_enforce, dialect_profiles, or verdict_stats. Usage is therefore inferable but not clearly directed.

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?

With no annotations, the description carries the full burden and discloses important behavior: resends may be refused without forwarding, and otherwise the tool classifies and records a verdict (a write side effect). It does not explain what a 'non-retryable verdict' is, what the refusal returns, or error conditions, which keeps it from a 5.

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?

A single compact sentence that front-loads the tool's role and then states the conditional behavior. It is dense but efficient, with no filler.

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 annotations and no output schema mean the description must carry all behavioral context. It explains the central refusal/classification mechanism but omits parameter meanings, return/error behavior, and the definition of a non-retryable verdict, leaving meaningful gaps for a tool with five required parameters and nested objects.

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 0% and six parameters exist. The description implies meaning for 'server', 'tool', and 'args' via '(server, tool, identical args)', but says nothing about 'mutating', 'prose', or 'structured'. The majority of parameters remain undocumented anywhere.

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 gives a specific verb+resource ('Gateway enforcement') and explains the core mechanism (refuse resend if a prior non-retryable verdict exists, otherwise classify and record). It is distinguishable from error_classify by its enforcement/refusal behavior, though it does not name that sibling explicitly.

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 conditional logic ('if a prior non-retryable verdict exists... otherwise classify + record') implies when the tool will refuse versus classify, but it never states when an agent should choose this tool over alternatives like error_classify or verdict_stats. Usage is inferable but not explicit.

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 exist, so the description carries the entire behavioral burden. It does not say whether this is a read-only aggregation, whether the counts cover a time window or full history, or how expensive/current the numbers are.

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?

It is short and front-loaded with the metric list, with no filler. However, it is an unpunctuated fragment without a leading verb or object, which reads like a field caption rather than a tool 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?

There is no output schema and no annotations, so the description is the only source of return-value information, and it gives only a partial field list. It omits the time scope, whether the counts are per-account cumulative, and any indication of response shape beyond the four named metrics.

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?

The tool takes zero parameters, so there is nothing for the schema or description to disambiguate; the baseline for a 0-param tool applies. The description does usefully enumerate which metrics are included in the payload.

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 the specific metrics (verdict mix, dialect hit rate, refused-resend, budget-stop counts), which tells an agent roughly what data this surface provides and differentiates it somewhat from error-oriented siblings. But it is a bare noun phrase with no verb or subject, so it never states what the tool actually does (retrieve? compute? aggregate?).

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?

No guidance on when to call this instead of dialect_profiles, error_classify, or error_enforce, and no stated preconditions or exclusions. An agent must infer from the metric names alone that this is a read-only summary path.

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