Skip to main content
Glama

concordance

by com.narrowhighway
OverviewDiscussionsAdmin

verify

[CHECK · door] Verify a claim deterministically — returns a verdict (HOLDS / PARTIAL / BROKEN / INCOMPLETE / SYSTEM_ERROR), the worked trail, AND a sealed receipt {content_hash, cite_url} you can re-fetch and re-verify (seal_fetch). Three forms: (a) PLAIN — {claim}: a plain-language claim; the engine finds the checkable claim(s) in it (de-identified in an airlock first), verifies and seals them, or answers INCOMPLETE if no extractor recognizes one; (b) MATH — {mode, params}; (c) ANY DOMAIN — pass steps, a list of {id, domain, spec} where spec is that domain's packet (e.g. {domain:'physics', spec:{PHYS_VERIFY:{mass_kg, acceleration_m_per_s2, claimed_force_N}}}). ~60 secular domains are covered (physics, medicine, finance, chemistry, ...); find_verifier(keyword) locates the right one. The engine eliminates what is not the answer; it does not generate it. READ THE VERDICT EXACTLY: only BROKEN is a finding about the claim. SYSTEM_ERROR means OUR verifier could not run (see means and error_at) and says NOTHING about whether the claim is true — never relay it to a human as a refutation. INCOMPLETE means no verifier applied (gap_at), or every claim found was governed by a negation, a reported belief or a condition and was DECLINED. PARTIAL means the checked fragment held and material text around it was NOT checked (coverage.unchecked_text) — never quote PARTIAL as HOLDS.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
modeNoMATH form: equality | inequality | derivative | integral | limit | solve
sealNomint a re-checkable seal (default true)
claimNoPLAIN form: a plain-language claim, in the caller's words
stepsNoDOMAIN form: [{id, domain, spec}] — spec is the domain's packet
domainNoPLAIN form: an optional domain hint (from find_verifier); recorded, never trusted to pick the verifier
paramsNoMATH form: e.g. {expr_a, expr_b, variables} for equality

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed2 schema fields changed
    • addedInput schema / properties / claim
      Added value: +{
+  "description": "PLAIN form: a plain-language claim, in the caller's words",
+  "maxLength": 4000,
+  "type": "string"
+}
    • addedInput schema / properties / domain
      Added value: +{
+  "description": "PLAIN form: an optional domain hint (from find_verifier); recorded, never trusted to pick the verifier",
+  "maxLength": 4000,
+  "type": "string"
+}
  2. Changed1 schema field changed
    • changedInput schema / properties / mode / enum
      Previous value: -[
-  "derivative",
-  "equality",
-  "inequality",
-  "integral",
-  "limit",
-  "solve"
-]New value: +[
+  "calc_chain",
+  "derivative",
+  "equality",
+  "inequality",
+  "integral",
+  "limit",
+  "number_theory",
+  "numeric",
+  "solve",
+  "system"
+]
  3. Changed2 schema fields changed
    • addedInput schema / properties / mode / enum
      Added value: +[
+  "derivative",
+  "equality",
+  "inequality",
+  "integral",
+  "limit",
+  "solve"
+]
    • removedInput schema / properties / mode / maxLength
      Removed value: -4000
  4. Changed5 schema fields changed
    • addedInput schema / additionalProperties
      Added value: +false
    • addedInput schema / properties / mode / maxLength
      Added value: +4000
    • addedInput schema / properties / params / additionalProperties
      Added value: +false
    • addedInput schema / properties / steps / items / additionalProperties
      Added value: +false
    • addedInput schema / properties / steps / maxItems
      Added value: +200
  5. First observed

TDQS

A4.6/5.0
Behavior5/5

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

With no annotations provided, the description carries the full burden and does so richly: it defines each verdict, warns that SYSTEM_ERROR says nothing about truth and must never be relayed as refutation, explains INCOMPLETE causes (no verifier, negation/belief/condition declined), and cautions that PARTIAL must never be quoted as HOLDS. It also discloses the sealed receipt {content_hash, cite_url} and its re-checkable nature via seal_fetch.

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?

The verdict/return contract is front-loaded and every clause conveys distinct operational value. It is information-dense and longer than typical, but each sentence earns its place; the parenthetical-heavy style is the only mild drag on readability.

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?

There is no output schema, so the description must explain return values — and it does: verdicts, worked trail, and sealed receipt, plus the semantics of ambiguous verdicts. For a 6-param nested-object tool with no annotations, this is complete enough for an agent to call and interpret results correctly.

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

Parameters4/5

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 genuine meaning beyond the schema: it explains the three parameter forms, gives a concrete DOMAIN spec example, and clarifies that the `domain` hint is 'recorded, never trusted to pick the verifier' — a subtlety not present in the schema text.

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?

States a specific verb+resource ('Verify a claim deterministically') and enumerates the exact verdict set returned, letting an agent distinguish it from siblings like attest_record or discern. It also declares the three input forms, so the resource and its scope are unambiguous.

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?

Explicitly maps input types to forms (PLAIN vs MATH vs DOMAIN) and names an alternative — find_verifier(keyword) — for locating the right domain packet. It does not, however, contrast itself against other verification/attestation siblings, so the when-to-use context is strong but not exhaustive.

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

All concordance toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources