corpus-grounding-expander

Ingest multi-version public-domain corpora into a unified verbatim grounding store.

7|Updated May 26, 2026
One-click install
npx skills add https://github.com/chrono-meta/forge-harness --skill corpus-grounding-expander
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: corpus-grounding-expander
Source: https://github.com/chrono-meta/forge-harness/tree/main/plugins/fh-meta/skills/corpus-grounding-expander
Command: npx skills add https://github.com/chrono-meta/forge-harness --skill corpus-grounding-expander

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves the problem of narrow, single-source grounding corpora that cannot support robust fail-closed verbatim relay for AI systems, leading to biased source verification and gaps in quote validation.

Core Features & Use Cases

  • Multi-Version Corpus Ingestion: Fetches and normalizes multiple public-domain versions of any verbatim source (religious texts, legal statutes, RFCs, technical standards) into a single unified key schema for cross-version quote alignment.
  • Provenance & License Tracking: Records literal per-version license, scope, and provenance metadata to ensure full legal compliance and full auditability of the grounding store.
  • Use Case: For teams building AI relay systems that require strict fail-closed verbatim grounding, this skill eliminates single-source bias by allowing quotes verified in any ingested version to count as valid, while ensuring no generated text is ever treated as grounded.

Quick Start

Use the corpus-grounding-expander skill to ingest 3 public-domain versions of the US Constitution as a unified verbatim grounding store for your compliance relay system.

Frequently Asked Questions about corpus-grounding-expander

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a multi-version verbatim corpus for strict AI source verification?

Build a verbatim corpus by ingesting and normalizing multiple public-domain document versions into a unified key schema. This creates a fail-closed grounding store that eliminates single-source bias during strict source verification.

What does fail-closed verbatim relay mean for AI grounding systems?

Fail-closed verbatim relay ensures no generated AI text is treated as grounded. Only quotes validated against the ingested multi-version corpus count as valid, preventing fabricated text from passing source verification.

Can I track license provenance across different versions of public-domain documents?

Yes, license provenance can be tracked across versions. The system records literal per-version license, scope, and provenance metadata during corpus ingestion to ensure full legal compliance and auditability of the grounding store.

How do I normalize cross-version keys for legal statute quoting and technical standards compliance?

Normalize cross-version keys by fetching multiple public-domain versions of legal statutes or technical standards and aligning them into a single unified key schema. This enables robust cross-version quote alignment for compliance relay systems.

What is the best way to prevent single-source bias in AI verbatim grounding corpora?

Prevent single-source bias by applying multi-version corpus ingestion. Validating quotes against any ingested public-domain version expands the grounding union safely while maintaining strict fail-closed fabrication probe specifications.

Does this grounding expander work with public-domain religious text analysis and document archival?

Yes, it supports religious text analysis and public-domain document archival. It ingests verbatim sources, applies cross-version key normalization, and validates quotes through fail-closed grounding union validation for accurate archival.