Skip to main content
Glama

extract_facts

Extracts hard data from chapter text and generates candidate fact contracts for fact_check review, flagging cross-chapter multi-value suspects. Candidates require human or LLM verification before use.

Instructions

【P1 闭环·零 LLM】从正文抽硬数据,生成候选事实契约(fact_check 兼容),含跨章同单位多值 suspect。候选 ≠ 确认:必须人工/LLM 复核后才能当正式契约。

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": {
      "chapters": "<chapters>"
    },
    "name": "extract_facts"
  }
}

Input Schema

NameRequiredDescriptionDefault
chaptersYes[{id,text}]

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.2.0

TDQS

A3.7/5.0
Behavior4/5

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

With no annotations at all, the description carries the burden and does substantial work: zero-LLM deterministic extraction, output is explicitly provisional rather than authoritative, and it predicts a specific tricky case (cross-chapter same-unit multi-value flagged as suspect). It still leaves auth, idempotency, and the exact shape of returned candidates unspecified.

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?

Compact and front-loaded: the mechanism, the output, the edge case, and the caveat in order, each clause doing work. The bracketed project jargon (P1 闭环, 零 LLM) is the only content that does not help an outside agent.

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 one-parameter, no-output-schema, no-annotation tool, this is nearly sufficient: the agent learns the input is chapter text, the result is unverified candidate facts, and one class of anomalies is pre-labelled. Only the returned structure and the review handoff remain unstated.

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?

A single required parameter (chapters array of {id,text}) at 100% schema coverage, so the schema already explains it; the description only implies 'body text' is the input. No added syntax, ordering, or size guidance beyond the schema, so this sits at the baseline.

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?

Names a specific verb (extract, 抽) and resource (hard data / facts from body text) and states the output artifact: candidate fact contracts compatible with fact_check. The scope (cross-chapter same-unit multi-value → suspect) further narrows it, though the sibling differentiation relies on the internal term fact_check rather than a named tool.

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

Usage Guidelines3/5

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

Implies usage: feed chapter text, get candidates, and the closing clause warns candidates ≠ confirmed so they must be reviewed before use as formal contracts. It never states when to prefer this over a sibling such as check_self_consistency, nor any prerequisite ordering, so guidance stays implicit.

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