Probe V0

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

Annotations already declare read-only, idempotent, and non-destructive behavior. The description adds valuable behavior beyond that: it is historical evidence lookup, does not make a live HTTP request, only works for an existing probe ID, and surfaces derived insights like counts and verdict changes. It stops short of describing the exact response format, but the provided behavioral context is strong.

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?

Three sentences with no fluff: core capability first, then specific use cases, then an explicit alternative. Every sentence adds distinct value and the key differentiator (no new HTTP request) is front-loaded.

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 read-only tool, the description covers what it returns conceptually, when it should be used, what it should be used after, and when it should not be used. Even without an output schema, an agent has enough context to invoke it correctly and interpret the results.

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%, so the description carries the explanatory burden. It contextualizes probe_id as an existing Probe experiment ID obtained after verify_http_claim and links the 'up to 20' cap to the limit parameter's maximum/default. It does not restate the hex format or allowed limit range, but those are already in the schema, and the usage context adds real 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?

The description uses a specific verb ('Retrieve'), a precise resource ('persisted observations for an existing Probe experiment ID'), and an explicit scope cap ('up to 20'), and immediately distinguishes it from the sibling by noting it makes no new HTTP request. This lets an agent recognize the tool and separate it from verify_http_claim without reading schemas.

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 an explicit trigger ('Use after verify_http_claim'), lists the inspection use cases (repeated results, PASS/FAIL/ERROR counts, agreement, verdict changes), and states the exclusion condition: use verify_http_claim when freshness or present endpoint behavior matters. Both when-to-use and when-not-to-use are unambiguous.

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 readOnly, openWorld, idempotent, and non-destructive; the description adds materially new behavior beyond them: bounded evidence persistence and reuse with freshness, manual redirect following capped at 3 validated hops with redirect: manual on every fetch, and the DNS rebinding caveat. This is rich, non-redundant disclosure that is fully consistent with the annotations.

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?

Roughly 115 words pack seven discrete facts — purpose, exclusions, assertion kinds, evidence behavior, redirect policy, URL constraints, security caveat — with no filler, and the purpose/exclusion sentence is front-loaded. 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 a tool with 5 parameters and a complex nested assertions schema, the description covers purpose, constraints, redirect policy, and evidence semantics thoroughly. However, there is no output schema and the description never specifies the result shape (e.g., per-assertion pass/fail, evidence payload, failure format), so an agent cannot anticipate how to consume the return value. That gap keeps it from a 5.

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 only 20% (only url carries a description), so the description carries the burden and largely delivers: it maps the four assertion kinds to their semantics (status, header_equals, json_equals at a JSON Pointer, body_contains), explains follow_redirects behavior, and implies freshness_seconds via 'reuses sufficiently fresh matching evidence'. It does not address how method=HEAD interacts with body_contains assertions or the 300–86400 freshness bounds, either of which would materially help an agent avoid failed calls.

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 names a specific verb and resource ('verify a current public HTTPS claim') and the method ('by direct observation'), explicitly excluding web search, documentation, browser inference, and shell/curl. The four assertion kinds (HTTP status, header value, JSON value at a JSON Pointer, body substring) make the scope precise and distinguish it from the probe.history sibling, which covers past observations.

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 use it — 'explicit assertions that a GET/HEAD currently returns' the listed expectations — and lists what it is not for (search, documentation, browser inference, shell/curl). The V0 URL restrictions (no credentials, query strings, fragments, IP literals, non-443 ports) also tell the agent when this tool cannot be used. The repeated emphasis on 'current' differentiates it from probe.history, though the sibling is not named explicitly.

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