field-notes

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

Annotations only declare the mute safety flags (readOnlyHint=false, destructiveHint=false, idempotentHint=false); the description supplies everything that actually matters operationally: the request is asynchronous, answered by an AI agent in roughly 2–10 minutes, free and unwarranted, rate-limited to 5/day, and the question plus context are retained after redaction of keys, tokens, JWTs, e-mails, IPs, and long encodings. The retention/redaction disclosure in particular is behavioral context the annotations cannot convey, and it is consistent with readOnlyHint=false since a ticket is created.

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 front-loaded correctly — purpose, then the follow-up tool, then behavior, then arguments, then limits, then examples — so an agent can stop reading early. It is a single run-on paragraph, though, and repeats the 4000-character limit already carried by the schema, which is the one place worth trimming.

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 compensates well by naming the ticket/status fields, the polling path via get_answer, expected latency, retention policy, and quota. Remaining gaps are minor: it does not say whether the ticket expires or what status values other than 'queued' can appear, which an agent polling get_answer would benefit from knowing.

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 and the description must earn more. It does add practical content guidance ('question or problem in plain words: what you tried, what you expected, what happened'; context as 'error text or logs'), but it also states 'max 4000 characters' for both question and context while the schema declares maxLength 40000 — a numeric mismatch that could actively mislead, which offsets the added value.

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 first clause states a concrete verb and resource ('Returns a ticket for a custom answer to your question') and immediately names the sibling that completes the flow ('fetch the answer with get_answer'). An agent can distinguish this from get_answer, find_cards, and get_card without opening a schema, and the sample output {"ticket":...,"status":"queued"} makes the return shape concrete.

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 explicit positive routing ('Good for \'my cron run is missing but the log shows no error\', \'how do I make a retry after an unknown broadcast idempotent?\'') and states the operational limit of 5 questions per agent per day. It does not tell the agent when NOT to use this tool — e.g. that existing answers should be sought with find_cards/get_card first — so a routing exclusion is missing.

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 only declare readOnly/idempotent/closed-world; the description adds substantial context beyond them: the cap of 3 results, the presence of source and checked_on fields, the issued ticket, redaction behavior, and that input is retained to improve the Field Guide. Those retention and redaction notes are exactly the kind of behavioral facts an agent cannot get from structured fields.

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-loads the return value and shape, then the argument, then privacy, then examples — a sensible order with no filler sentences. It is a dense single paragraph with long clauses, so it is slightly harder to scan than it needs to be, but every sentence carries information.

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 compensates by sketching the response (ticket plus up to 3 cards) and giving a literal example. Combined with the retention/redaction warning, an agent has enough to call it correctly; the only residual confusion is the 2000 vs 20000 character limit mismatch.

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, but the description adds real meaning: what counts as a valid value (exact error message or one-line description), the 2000-character intent, and the privacy consequence of what is pasted. Minor flaw: it says 'at most 2000 characters' while the schema's maxLength is 20000, which could mislead on the true limit.

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 ('Returns up to 3 checked field notes that match an error text') and enumerates the three card categories it can return (pitfall cards, deprecated models/API versions, non-browser login routes). It also distinguishes itself from the sibling report_outcome by explaining that the ticket in the response feeds 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?

Gives concrete positive examples of when it applies ('insufficient_quota', 'invalid_grant', 'spawn npx ENOENT', 'my list returned 30 items, is that all?'), which is strong signal. It does not explicitly exclude cases or say when to prefer get_card or get_answer instead, so routing versus siblings is only implied.

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 cover the safety profile (readOnlyHint, idempotentHint, openWorldHint=false), and the description adds real value on top: the full state machine, that queued includes a position, and the caveat that 'the answer is written by an AI agent and provided as is'. That caveat is exactly the kind of trust/reliability context annotations cannot express.

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 state enumeration, and the inline example JSON makes the answered shape concrete. The 'Argument: ticket.' sentence earns little and the AI-agent caveat sits at the tail where it is easy to skim past, but overall there is minimal waste.

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 carries the return-value burden and does so via explicit states plus a sample payload. Remaining gaps are operational — polling cadence, ticket expiry, and what to do on 'failed' — which are minor for a single-parameter read 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 coverage is 100% and the single parameter is already documented as 'Ticket from ask_markedet.' The description's line 'Argument: ticket.' restates the schema rather than adding format, lifetime, or reuse semantics. Baseline 3 applies since the schema carries the parameter meaning.

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 ('Returns the status of an ask_markedet ticket') and enumerates the exact return states — queued with position, answering, answered with answer and field notes, failed. Referencing ask_markedet ties it to the sibling that creates tickets, so an agent can distinguish it from get_card or find_cards without opening a 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?

The tie to ask_markedet and the 'queued (with position)' state make the polling-after-submission pattern clear, and the 'ticket' argument is anchored to the tool that produces it. It stops short of explicit guidance such as when to poll versus when to stop, or what to do on 'failed', so no exclusions are stated.

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 cover the safety profile (readOnly, idempotent, closed-world), but the description adds the return shape with a sample card object, which is valuable since no output schema exists. Minor tension: it claims the note is returned "in full" yet the sample shows only kind/id/title, leaving field completeness slightly ambiguous.

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?

Purpose, identifier formats and a return example are packed into two tight sentences with the core action front-loaded. The bare "Argument: id." fragment is redundant filler next to the schema and example.

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 single-parameter lookup with no output schema, the description supplies enough: identity of the resource, id formats, and a sample response. Only the full set of returned fields is left unstated, which the "in full" claim partly obscures.

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% and the schema's id field already lists the PC-001 / DP-0064 / NH-001 examples, so the description's prefix list and "Argument: id" line add little beyond it. Baseline 3 applies because the schema carries parameter semantics.

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 ("Returns one field note in full by id") and pins down the identifier namespaces (PC-, DP-, NH-). The example output makes the returned object concrete, so an agent can distinguish this from find_cards 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?

Gives a clear usage context: "Good for reading the snippet, cause and source excerpt of a card that find_cards returned," which positions it as the follow-up to a search call. It stops short of explicit when-not guidance or naming find_cards as the retrieval alternative for lookup-by-topic.

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?

Beyond the annotations (write, non-idempotent, non-destructive, closed-world) the description discloses the exact return shape, the follow-up mechanic, and the privacy consequence of omitting a handle ('without a handle only an anonymous count per card is kept'). It also rules out a free-text field so the agent doesn't waste a call attempting narrative feedback, and states the character restriction on handle/platform.

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?

Purpose and return value are front-loaded, and the 'Arguments:' block is dense but organized by parameter with no filler. It reads as a run-on paragraph with telegraphic phrasing, slightly harder to scan than a short structured list would be.

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 six-parameter mutation tool with no output schema, it covers what is written, what is returned, and the identity/privacy tradeoff, which is most of what an agent needs. It does not mention failure behavior for an invalid or expired ticket, and does not say whether re-reporting the same ticket overwrites a prior outcome — relevant given idempotentHint=false.

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 67% and two enum parameters carry no descriptions there; the description supplies their meaning (worked/did_not_work/partly, yes/no/unsure) plus the purpose and format constraint for agent_handle and platform. The remaining gap is that it doesn't say when card_id matters versus when the ticket suffices, so one optional param is still ambiguous.

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 first sentence states a specific verb and resource ('Records whether a fix ... worked') and anchors it to the two sibling tools that produce the ticket (find_cards, ask_markedet). An agent can immediately tell this is the feedback/verification sink for those siblings rather than a lookup tool.

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 a concrete trigger ('Good for "the verify step passed"') and an implicit sequencing rule — call this after find_cards or ask_markedet produced a ticket. It also frames the tradeoff that governs the optional parameters: supply a handle if you want confirmation, omit it for anonymous counting. It stops short of explicit when-not conditions (e.g., don't report on a ticket you never used).

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