describe_what_i_wantTurn a plain sentence ('remote data roles in Germany or the Netherlands, senior, not agencies') plus the saved CV into concrete targeting rules. Returns the inferred rules and the chips a person would confirm; with apply=true they are saved and the feed re-matches. preset: balanced | broad | focused.
| Name | Required | Description | Default |
|---|---|---|---|
| apply | No | ||
| preset | No | balanced | |
| sentence | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations declare readOnlyHint=false, and the description correctly says apply=true saves and re-matches the feed, so mutating behavior is disclosed. It also mentions the non-apply path returns inferred rules and chips for confirmation. With no openWorldHint/idempotentHint help, these details add meaningful behavioral context. A slightly higher score is constrained by not describing what happens to the saved CV or whether it is used automatically.
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 packs a lot of information into a compact, front-loaded structure. The first sentence states purpose; the second describes outputs and mutation behavior; the third enumerates preset options. No unnecessary words or repeated schema 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?
For a tool with 3 simple parameters and no output schema, the description provides enough to call it correctly: input format, preset options, apply behavior, and what the response contains. Missing is a caveat about maxLength or what happens with an empty/ambiguous sentence, but that is minor and inferable. The sibling landscape is large, and the description still makes this tool identifiable.
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 must compensate. It explains sentence with a concrete example, explains preset via its enum values, and explains apply via its behavioral effect. That covers all three parameters meaningfully, which is strong for a description that is only two sentences long.
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 clearly states the tool's purpose: turning a plain sentence plus the saved CV into concrete targeting rules. It also distinguishes itself from siblings like update_targeting by focusing on natural-language parsing rather than direct targeting edits. The verb 'turn' with the resource ('concrete targeting rules') is specific, and the example sentence makes the input format immediately understandable.
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 conveys when to use it: when a user expresses targeting preferences in a plain sentence and wants them converted into structured rules. It does not explicitly name alternatives like update_targeting, but the distinction is implicit through 'plain sentence' vs direct editing. The apply=true mention clarifies the save-versus-preview behavior, which is useful usage guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.
Tool purposes are generally distinct and well-described, but a few clusters overlap in function: answer_screening_question vs save_answer both write to the answer book, get_my_profile vs get_account both report plan status, and the CV preview/sent-CV/base-CV tools could be confused. The detailed descriptions mitigate most misselection, so this is only a minor issue.
The set almost uniformly uses snake_case verb_noun names like list_, get_, update_, create_, delete_, and start_/stop_. Minor deviations such as login, describe_what_i_want, and the get_my_* vs list_* alternation prevent a perfect score, but the overall pattern is predictable and readable.
49 tools is far above the 25+ threshold and will burden agent tool selection even though many are legitimate single-purpose operations. Several groups could be consolidated—billing links, API-key management, and the CV PDF family—without hurting clarity.
The surface covers the full lifecycle: account creation/auth, profile and CV, targeting, matching, apply runs, screening answers, tracking, billing, export, and deletion. Minor gaps remain, such as no application-level detail/withdrawal endpoint and no direct way to save a parsed CV without re-uploading, but agents can work around them.