Skip to main content
Glama

Knowledge grounding check

wever_knowledge-grounding-check
Read-only

Agent Knowledge: submit one answer with the sources it cites: optionally the question and the answer as given; each claim the answer makes with its id, its text and the ids of the sources it cites; and each cited source with its id, its title and the cited passage as the source states it. The signed report reads claim by claim whether a cited passage makes the claim, stating every figure in it and at least 60% of its content words read without case, stop words or word endings, lists every citation with the figures and words it shares and lacks, and names the claims the answer cannot stand on: claims that cite no source, and citations attached to claims their passages do not make, such as a figure the passage never states; then claims stronger than their source, which state as always, never, all or guaranteed what the passage hedges, or state flat a figure the passage gives as a ceiling such as up to 40%, and claims that also cite passages that do not make them, with the fix for each. A cited source supplied with no passage is named as an open point and never fetched. One product or diligence claim against pasted evidence with no citations is wever_claim-evidence-check, a benchmark figure against its source is wever_benchmark-evidence-check and whether the sources behind a knowledge base can be checked at all is wever_knowledge-source-check. Verdict answer_ungrounded, grounding_gaps, grounding_unverified or answer_grounded. POST /api/knowledge-grounding-check. Reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write. Paid URL: 0.10 USDC per call on Base via x402 exact, with the shared 10 signed-wallet free calls per rolling 30 days. Bind X-Wever-Wallet-Proof to the exact JSON arguments and the tool route. Returns a report with an Ed25519 signed receipt. Contract: /api/agent-knowledge-contract. Human page: https://weverlabs.com/agent-knowledge/.

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": {
      "claims": "<claims>"
    },
    "name": "wever_knowledge-grounding-check"
  }
}

Input Schema

NameRequiredDescriptionDefault
labelNoOptional name recorded in the report.
answerNo
claimsYesEach claim the answer makes.
sourcesNoEach cited source with its passage.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A3.6/5.0
Behavior4/5

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

Annotations already declare readOnlyHint=true, openWorldHint=false and destructiveHint=false, and the description reinforces this with 'reads only the material in the call: no network request, fetch, money movement, message, sharing, configuration change or external write.' It adds behavior the annotations cannot carry: the 60% content-word/figures matching rule, the open-point handling for sources supplied without a passage, the verdict enum, the x402 pricing (0.10 USDC on Base, 10 signed-wallet free calls per rolling 30 days), the X-Wever-Wallet-Proof binding requirement, and the Ed25519 signed receipt. Missing is whether the call is idempotent or how repeat submissions are billed, which the annotations do not answer either.

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

Conciseness2/5

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

The core sentence is a single run-on of roughly 150 words that packs the matching rule, the report structure, three failure taxonomies, the fix output and two verdict branches into one breath; the reader must parse 'stating every figure in it and at least 60% of its content words read without case, stop words or word endings' mid-sentence. The routing and pricing sentences are cleanly front-loaded, but the payload description is not structured into scannable segments and repeats boilerplate (contract URL, human page, route) that crowds out the semantics an agent needs. Substantial trimming or bullet-style segmentation would materially improve it.

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

Completeness4/5

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

For a four-parameter, nested-object mutation-free check with no output schema, the description covers what the report contains (claim-by-claim grounding, shared/lacking figures and words, the ungrounded/overclaim categories, per-claim fixes, open points) and the verdict set, so an agent knows what it will receive. It also covers the billing and signing prerequisites needed to invoke it successfully. The remaining gap is that with no output schema the exact report shape and receipt fields are only described narratively, leaving ambiguity about how to parse the response.

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

Parameters3/5

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

Schema coverage is 75%, so nested fields like label, answer.text/question and cites are already documented in the schema. The description adds real semantic constraints the schema does not: sources must carry the passage as the source states it and a source without a passage becomes an open point rather than a fetched one, and claims' cites are matched against those supplied passages. However, it never explains the meaning of the claim id versus source id namespace, the 200/500 caps, or what happens when a cited id has no matching source entry, so it only partially compensates for the remaining coverage 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 specific verb (submit one answer with the sources it cites) and resource (claims plus cited passages) and explains what the signed report adjudicates claim by claim. It distinguishes itself from siblings by naming wever_claim-evidence-check, wever_benchmark-evidence-check and wever_knowledge-source-check with the conditions that select them. The opening 'Agent Knowledge: submit one answer with the sources it cites' is somewhat opaque as a purpose statement, so it is clear rather than crisp.

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 explicitly routes three adjacent cases elsewhere: one product/diligence claim against pasted evidence with no citations goes to wever_claim-evidence-check, a benchmark figure against its source to wever_benchmark-evidence-check, and whether sources behind a knowledge base can be checked at all to wever_knowledge-source-check. It also discloses the no-citation case is handled as an open point rather than fetched. It stops short of stating when NOT to use this tool for multi-claim/multi-source bundles, but the routing is clear for the sibling set.

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