Ai Model Experiments

Ask Pipeworx — Grounded

ask_pipeworx_grounded

Read-onlyIdempotent

Hallucination-resistant answer mode for high-stakes reads. Same routing as ask_pipeworx — picks the right tool from 7,231 across 1876 sources, fills arguments, fetches the data — then EXTRACTS the answer using ONLY what the tool result contains. Returns {answer, evidence (verbatim quote), confidence, source, fetched_at, refusal_reason:null} on success, OR an explicit refusal {answer:null, refusal_reason:"not_in_source"|"no_tool_match"|"tool_error"|"data_truncated"|"llm_error"} when the data doesn't directly answer. An "llm_error" also carries refusal_cause ("wall" = our time budget ended before any model call was made; "route"|"threw"|"unparseable"|"wall_after_failure" = a model call failed); refusal_reason itself is unchanged. Use whenever an answer will be quoted, cited, or acted on, and the agent must not invent facts (financial verdicts, legal claims, medical lookups, public statements). Costs one extra LLM call vs ask_pipeworx — prefer ask_pipeworx for casual lookups.

Example tool call

This call uses example arguments supplied in the tool’s input schema.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "question": "What was Apple's fiscal 2023 revenue?"
    },
    "name": "ask_pipeworx_grounded"
  }
}

Input Schema

Name	Required	Description
`q`	No	Alias for question.
`ask`	No	Alias for question.
`text`	No	Alias for question.
`input`	No	Alias for question.
`query`	No	Alias for question.
`prompt`	No	Alias for question.
`message`	No	Alias for question.
`question`	Yes	Your question in natural language. Accepts query, q, prompt, text, input, ask, message as aliases.
`prefer_tool`	No	Run this exact tool instead of letting the router choose — use a name from a previous answer's alternatives[]. Arguments are still filled from your question. Unknown names are refused, never guessed.

Schema Changelog

Changes observed during successful MCP inspections.

Changed5 schema fields changedOct 19, 2026

addedInput schema / examples

Added value: +[
+  {
+    "question": "What was Apple's fiscal 2023 revenue?"
+  }
+]

addedInput schema / properties / ask

Added value: +{
+  "description": "Alias for question.",
+  "type": "string"
+}

addedInput schema / properties / message

Added value: +{
+  "description": "Alias for question.",
+  "type": "string"
+}

addedInput schema / properties / prefer_tool

Added value: +{
+  "description": "Run this exact tool instead of letting the router choose — use a name from a previous answer's alternatives[]. Arguments are still filled from your question. Unknown names are refused, never guessed.",
+  "type": "string"
+}

changedInput schema / properties / question / description

Previous value: -"Your question in natural language. Accepts query, q, prompt, text, input as aliases."New value: +"Your question in natural language. Accepts query, q, prompt, text, input, ask, message as aliases."

First observedSep 11, 2026

TDQS

A4.6/5.0

Behavior5/5

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

Annotations only cover the safety profile (read-only, open-world, idempotent), while the description adds substantial behavior beyond them: the success payload shape, a full refusal taxonomy with five reason codes, the nested refusal_cause semantics for llm_error, and the cost tradeoff of one extra LLM call. This is exactly the context annotations cannot convey.

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?

Front-loaded with the value proposition, then the return contract, then usage guidance — a logical order. The middle sentences are dense with enumerated refusal codes, but each item carries distinct information; nothing is filler.

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?

No output schema exists, and the description compensates by specifying both the success object and the refusal object shapes in full. For a tool whose distinguishing risk is hallucination and failure modes, every case an agent must handle is described.

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 100% and the alias set (q/ask/text/input/query/prompt/message) is already documented inline, so the description adds little parameter-level detail. It only hints that arguments are 'filled from your question' and does not mention prefer_tool or its 'never guessed' refusal behavior, which the schema covers.

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 and mode: routing like ask_pipeworx, then EXTRACTING the answer using only what the tool result contains. It explicitly distinguishes itself from the sibling ask_pipeworx ('same routing as ask_pipeworx... costs one extra LLM call'), so an agent can pick between them without opening either schema.

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

Usage Guidelines5/5

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

Gives an explicit when-to-use ('whenever an answer will be quoted, cited, or acted on... must not invent facts' with concrete domains: financial verdicts, legal claims, medical lookups, public statements) and an explicit when-not-to ('prefer ask_pipeworx for casual lookups'). Nothing is left to inference.

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

All Ai Model Experiments tools Inspect

Try in Browser

Claude Code, inside the Godot 4 editorFlockbay puts an AI agent in the real Godot 4 editor: it writes GDScript, runs your game and fixes what breaks. Free, or bring your own Claude Code or Codex.flockbay.comAd

Glama MCP Gateway

Add one secure layer between your agents and this server.