Skip to main content
Glama

Check this first

agency_check
Read-onlyIdempotent

Get a second look before an agent buys, sends, books, shares, signs up, starts a trial or subscription, connects an app, or says a task is done. Call it whenever the person asks whether to go ahead with any of these, including questions like "should it go ahead?" or "is this a good idea?", before you give your own opinion. Show the card first, then add your own analysis below it. Returns one Decision Card: Ready to review, Clarify one thing, or Pause and verify, with what is supported, what is not yet supported, what is unknown, and one next step. It reviews only what is sent to it. It never contacts anyone, blocks anything, or acts. If the card asks one question, call again with the same request plus answers: {rule_id: option_key}.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "context": "<context>"
    },
    "name": "agency_check"
  }
}

Input Schema

NameRequiredDescriptionDefault
actionNo
claimsNo
localeNo
schemaNo
senderNo
answersNoAnswers to a prior Clarify card, keyed by the rule that asked.
contextYes
sourcesNo
memoriesNo
user_goalNo
local_timeNoUser's local clock time when the check was asked for. Supplied by the caller, never read from the server clock.
request_idNo
form_fieldsNo
invoked_viaNo
action_recordNo
ruleset_versionNoOptional pin. The engine refuses to evaluate under any other version.
completion_evidenceNo

Output Schema

NameRequiredDescriptionDefault
cardYes
receipt_idYes
ruleset_versionYes

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A3.5/5.0
Behavior4/5

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

Annotations already declare readOnlyHint/idempotentHint true and destructiveHint/openWorldHint false, so the safety bar is low, yet the description still adds real value by naming the three possible card outcomes, listing the card's contents, and clarifying 'it never contacts anyone, blocks anything, or acts' — disambiguating that this is advisory rather than a gate. It also discloses the retry protocol for Clarify cards. Return-value detail is partially redundant with the existing output schema, but the non-obvious 'does not block' behavior is worth the sentence.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Sentences are purposeful and mostly one job each: purpose list, invocation trigger, display order, return shape, non-actions, retry. The opening enumeration ('buys, sends, books, shares, signs up, starts a trial or subscription, connects an app, or says a task is done') is longer than needed and slightly buries the core verb. No filler otherwise.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness2/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

This is a high-complexity tool with 17 parameters, deeply nested objects, and rich enums, and the description addresses only the return shape and the retry loop. An agent reading the description alone would not know what to put into context, action, sources, memories, or form_fields, which are the substance of the check. The output schema existing excuses return-value documentation, but not the input-side omission.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters2/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With 17 parameters at 18% schema description coverage, the description carries a heavy compensation burden and covers essentially one field: 'answers' keyed by rule with option_key values. The large payloads an agent must populate — action.kind, terms, claims, memories, form_fields, sender, user_goal, completion_evidence — get no mention or meaning in the description. Only the sparse top-level schema description ('share only what the check needs... codes, card numbers and passwords are removed') partially fills the gap.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a concrete function: a second look before an agent commits to an action (buying, sending, booking, signing up, installing a connector) and returning one Decision Card. The verb+resource are specific enough that an agent can tell it apart from describe_ruleset, explain_rule, or get_receipt without opening the schema. It stops short of 5 only because it never names those siblings or states the boundary against them.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Trigger conditions are unusually explicit: 'call it whenever the person asks whether to go ahead,' including the phrasing 'should it go ahead?' and 'is this a good idea?', plus the sequencing rule 'show the card first, then add your own analysis below it.' It also documents a re-invocation path when the card asks a question. The gap is that it gives no explicit when-NOT-to-use case or named alternative tool.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.