Skip to main content
Glama

beatandraise.com SEC Filings

Check Citations in a Draft

check_citations
Read-onlyIdempotent

Check every number in a draft against the filing characters it cites, and link them.

Write your draft with a citation right after each figure, using the ref that list_figures or verify_figure gave for it: revenue of $68.1 billion [ref:0001045810-26-000019/q4fy26pr.htm@112+4] For a number you calculated, state the formula, then cite its inputs; a, b, c… are the refs in the order you cite them: up 73% [calc: a/b-1, ref:, ref:] free cash flow margin 53.9% [calc: (a-b)/c, ref:, ref:, ref:] Formulas use a-h, + - * /, parentheses and constants (a4, a/b100). A measure such as a margin can be defined more than one way, so the formula you state is the one checked and shown (formula_source "stated"); a figure printed in parentheses is tried as negative, then as a positive amount (signs). Without a formula, the usual sum, difference, ratio and growth rate are tried (formula_source "inferred"). You may also give a calc's value: [calc: 53.9% = (a-b)/c, ref:…]; without it, a calc is for the number just before the bracket. A ref is for the number in its sentence that equals the characters it points at, wherever it sits, so a ref at the end of a sentence works, and a year is checked like any figure when the ref points at a year. A ref names the exact characters in the document: /@+.

Returns per citation: ok / rounded (your number is the cited figure times a power of ten, rounded), mismatch (the cited characters, or your formula of them, give a different number), bad_ref (no number at that position), bad_formula (unreadable, or not using every cited input), bad_claim (a calc's stated value is not in its sentence), not_derivable (no formula stated, and no sum, difference, ratio or growth rate of the inputs gives your number). unsourced_numbers lists figures with no citation. Fix every failure before presenting the text, then use linked_text and footnotes: each footnote links to the highlighted figure in the filing, so a reader can check it.

Args: text: The draft, with [ref:…] / [calc: …] citations (max ~200 KB)

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": {
      "text": "<text>"
    },
    "name": "check_citations"
  }
}

Input Schema

NameRequiredDescriptionDefault
textYes

Output Schema

NameRequiredDescriptionDefault

No arguments

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.5/5.0
Behavior4/5

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

Annotations already declare readOnlyHint/idempotent/openWorld, so safety is covered; the description adds genuinely new behavior — the full status taxonomy (ok, rounded, mismatch, bad_ref, bad_formula, bad_claim, not_derivable), the unsourced_numbers report, the linked_text/footnotes output, the ~200 KB size cap, and the inferred-vs-stated formula_source distinction. Some of this overlaps the output schema, but the meaning of each failure mode is not something an agent could infer from hints alone.

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?

Purpose is front-loaded and the worked examples anchor the syntax well, but the run of prose rules ('You may also give a calc's value… A ref is for the number in its sentence…') is dense and could be broken into tighter bullets. Nearly every sentence carries real specification content, so the length is defensible rather than padded.

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

Completeness5/5

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

For a genuinely complex parsing/verification tool with a one-field schema, the description supplies the input grammar, formula grammar, all failure semantics, the size limit, and what to do with the result. With an output schema present and read-only annotations set, nothing essential an agent needs to call this correctly is missing.

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

Parameters5/5

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

Schema coverage is 0% — a single bare 'text' string — so the description carries the entire burden, and it does: it specifies the exact citation grammar [ref:<accession>/<filename>@<start>+<len>], the calc syntax with named inputs a–h, the allowed operators and constants, optional stated values, and how a ref binds to the number in its sentence. Without this the parameter would be unusable.

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

Purpose5/5

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

The opening sentence gives a specific verb and resource — 'Check every number in a draft against the filing characters it cites, and link them' — which clearly separates it from sibling verify_figure (single figure) and verify_quote. The citation-ref format echoed from list_figures/verify_figure reinforces how it differs from those tools.

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?

It states the preconditions (write the draft with a [ref:…] or [calc: …] citation right after each figure), names the sibling tools that supply the refs, and prescribes the post-step ('Fix every failure before presenting the text, then use linked_text and footnotes'). It never explicitly says when NOT to use it, e.g. to check a single figure with verify_figure instead, so it falls short of a 5.

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.

Resources