Skip to content
Back to skills

Claim Ledger Skill

ASecurity

Normalize collected observations into auditable claims with support, contradiction, provenance, limitations, and confidence.

  • 3 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 2, 2026
ai-agents

Security analysis

A100/100

Scanned September 2, 2026

npx -y skills add kinqsradiollc/BrainRouter --skill claim-ledger-skill --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Claim Ledger Skill?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Claim Ledger Skill
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/kinqsradiollc-claim-ledger-skill/badge)](https://www.skillsdirectory.com/skills/kinqsradiollc-claim-ledger-skill)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: claim-ledger-skill
description: Normalize collected observations into auditable claims with support, contradiction, provenance, limitations, and confidence.
allowed-tools: [read_file, list_dir, grep_search, glob_files, research_note, research_brief]
---

# Claim ledger

## Overview

The claim ledger is the boundary between collection and synthesis. It prevents
source excerpts, analyst interpretation, and final conclusions from being
blurred together and makes conflicts and evidence gaps inspectable.

## When to Use

Use after collecting evidence, during long investigations, or whenever multiple
sources may support, qualify, or contradict the same material claim.

## Workflow

1. Convert each material observation into the smallest independently testable
   claim. Keep observed source content separate from analyst inference.
2. Attach stable source identity, location, date, relation to the claim
   (supports, refutes, qualifies, or unclear), and relevant limitations.
3. Deduplicate equivalent claims while preserving independent provenance.
   Related passages from one underlying source count as one evidence lineage.
4. Assign confidence from evidence quality, relevance, independence, and
   agreement. Do not infer confidence from source count alone.
5. Link each claim to its sub-question and mark unsupported, single-source,
   stale, or conflicting claims for follow-up.
6. Freeze a reviewable ledger snapshot before synthesis. New evidence after
   that point creates an explicit revision rather than silently changing claims.

## Verification

- [ ] Every material claim is atomic and linked to a sub-question.
- [ ] Source fact, interpretation, and inference are visibly distinct.
- [ ] Evidence lineages are deduplicated and their relationships are labelled.
- [ ] Confidence and open conflicts have an inspectable basis.

## Red Flags

- Copying whole documents into the ledger instead of extracting claims.
- Counting syndicated or repeated material as independent corroboration.
- Raising confidence because a claim sounds plausible.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.