SUPTOC is a deterministic auxiliary tool for LLM products. It processes TOC structure instead of freely generating content.

Why does it avoid hallucination?

The system is constrained by input text and human language rules. It extracts, detects, validates, and corrects rule-level issues without inventing missing headings.

How is API access priced?

API access requires membership. Each membership level costs 1,000 CNY and includes 100,000 API calls.

Deterministic TOC assistant for large-model workflows

Millisecond TOC structure processing without hallucination

SUPTOC is an auxiliary tool for LLM applications. It extracts, detects, validates, and corrects table-of-contents structures by strict human language rules instead of inventing headings that do not exist.

View member API

msprocessing speed

0hallucinated titles

10万calls / level

Done in milliseconds

LLM tool call

0invented headings

100%source-bound

mslatency

Structured TOC

no hallucination

L1Chapter 1 Overview
p.1100%
- L21.1 Product Positioning
  p.1100%
- L21.2 Core Rules
  p.2100%
  - L31.2.1 Millisecond Processing
    p.2100%
- L21.3 No Hallucination Output
  p.3100%
L1Chapter 3 API Access
p.6100%
- L23.1 Membership Activation
  p.6100%
- L23.3 API Usage Quota
  p.7100%

ExtractDetect styleValidateCorrect

Built for LLM apps, RAG systems, editors, document platforms, and knowledge teamsLLM pluginsRAG ingestionDocument editorsKnowledge systems

Core capabilities for reliable TOC structure work

SUPTOC does not replace the LLM; it gives the LLM a stable, fast, and verifiable TOC tool layer.

↳

Extract text TOC structure

Extract hierarchy, parent-child relations, and heading paths from plain text, model output, or document TOCs.

◇

Detect current TOC style

Recognize numeric numbering, Chinese chapters, English Chapter labels, indentation, and mixed styles.

✓

Validate structure issues

Find skipped numbering, missing levels, duplicate headings, hierarchy drift, and style inconsistency.

✎

Auto-correct

Fix numbering and hierarchy by human language rules without inventing headings not present in the input.

Millisecond speed

Rule-first processing makes it suitable as an LLM function call and tool layer.

No hallucination output

Results are constrained by the source text for reliable automated workflows.

A deterministic pipeline for LLM assistants

Input TOC

Receive text, document TOC, or LLM output.

Rule parsing

Parse by language, numbering, indentation, and style rules.

Validate

Check skips, missing levels, duplicates, and drift.

Correct

Return structured TOC and explainable issue reports.

Designed as a stable tool layer for large models

▣

LLM function calling

Let the LLM call SUPTOC for deterministic TOC structure.

▣

RAG document ingestion

Validate hierarchy before ingestion to reduce bad context.

▣

Document editors

Detect numbering and hierarchy issues automatically.

▣

Content QA

Batch-check manuals, reports, policies, and course materials.

API

Member-only API with 100,000 calls per level

API access requires membership. Each 1,000 CNY level includes 100,000 API calls for LLM apps, editors, and document automation systems.

View API page

curl -X POST https://api.suptoc.com/v1/toc/validate-and-correct \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"Chapter 1 Overview\n1.1 Product\nChapter 3 API"}'

Rule-bound processing, not free-form generation

CapabilityGeneric LLMSUPTOC

TOC extractionMay rewrite or inventSource-bound

Structure validationUnstableExplainable rules

Processing speedModel-latency boundMilliseconds

Correction boundaryMay over-generateOnly rule-level fixes