Skip to content
Back to skills

Evidence Chain Verification Lsdisconzi

ASecurity

Verify the authenticity, chain of custody, and forensic integrity of all evidence items. Check hashes, source-to-transcript alignment, speaker diarization accuracy, and identify gaps requiring preservation orders.

  • 85 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 8, 2026
ai-agentsrustgorailsgit

Works with

  • cli

Security analysis

A100/100

Scanned October 8, 2026

npx -y skills add ThomasMoreAI/legal-skills-open --skill evidence-chain-verification-lsdisconzi --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Evidence Chain Verification Lsdisconzi?

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

Security grade badge for Evidence Chain Verification Lsdisconzi
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/thomasmoreai-evidence-chain-verification-lsdisconzi/badge)](https://www.skillsdirectory.com/skills/thomasmoreai-evidence-chain-verification-lsdisconzi)

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

SKILL.md
---
name: evidence-chain-verification-lsdisconzi
title: Evidence Chain Verification
description: Verify the authenticity, chain of custody, and forensic integrity of all evidence items. Check hashes, source-to-transcript alignment, speaker diarization accuracy, and identify gaps requiring preservation orders.
author: lsdisconzi
author_url: https://github.com/lsdisconzi/craudio-para-adevogados/tree/main/skills/evidence-chain-verification
license: Apache-2.0
version: 0.1.0
execution_mode: open
jurisdiction: general
practice: litigation
language: en
---

# Evidence Chain Verification

Systematic verification of the evidentiary record's forensic integrity. Ensures every piece of evidence is authenticated, hashed, and chain-of-custody documented before use in any filing. This is a deterministic engine — it names no party or case; it operates over whatever evidence registry the active matter declares.

## Running this skill

1. **Load configuration.** Read the configured matter profile and firm profile from the plugin config directory (see CLAUDE.md `## Configuration Location`). If a file is missing, or a critical field still shows `[PLACEHOLDER]`, STOP — tell the attorney: "This plugin needs setup. Run the `cold-start-interview` skill."
2. **Apply the frame.** The Plugin Operating Rules, Shared Guardrails, Decision posture, Jurisdiction recognition, Retrieved-content trust, and Ontology Governance sections of CLAUDE.md govern this skill. The workflow below is a FLOOR, not a ceiling.
3. **Resolve vault paths.** Every `{vault root}` reference resolves against `## Vault location` in CLAUDE.md. Never hard-code an absolute path.
4. **Apply the work-product header** from CLAUDE.md `## Outputs` to every internal deliverable; suppress it on externally-facing output per Quiet mode.
5. **Run the workflow below.**
6. **Close** with the `⚠️ Reviewer note` block and the next-steps decision tree from CLAUDE.md `## Outputs`. Run the `ontology-validate` skill on the output before it reaches a court or counterparty.

---

## Workflow

### 1. Inventory check
- Load every evidence item from `{vault root}/03-evidence/`.
- Cross-reference with the transcript note map at `{vault root}/02-transcripts/transcript_note_map.json`.
- Flag any evidence item not found in the vault at its declared path.

### 2. Hash verification
- Items marked `UNHASHED` or `UNHASHED_ATTACHMENT_REF`: generate a SHA-256 hash.
- Items marked `INDEX_MAP_VERIFIED`: confirm the hash matches the registry.
- Items with missing source files: flag for client/curator action.
- **Command:** `sha256sum <path>` on all media files.

### 3. Transcript-to-source alignment
For each transcript with a media source, verify:
- Duration matches between transcript metadata and the media file
- Speaker count is consistent
- Segment timestamps fall within the media duration
- No unexplained gaps longer than 10 seconds
- Flag any transcript without a verified media source.

### 4. Chain of custody
For each evidence item, verify the custody chain is documented: who recorded it, when, where it is stored, who has accessed it (curation log), and that no modifications were made to original evidence. Flag any break in the chain.

### 5. Speaker identification
- Cross-reference speaker labels across transcripts.
- Map `SPEAKER_XX` labels to named actors where identified.
- Flag transcripts where critical speakers are not yet identified.

### 6. Gap analysis
Identify evidence gaps requiring action. For each gap, record: what is missing, why it matters to the case theory, the jurisdiction-appropriate mechanism to obtain it (preservation order, information request, discovery request), and the deadline pressure. Common gap classes: counterparty-held recordings, official records held by a public body, internal counterparty documents, personnel records of named actors.

### 7. Output
- **Evidence integrity report** — per-item hash and chain status
- **Gap matrix** — what is missing, why it matters, how to obtain it
- **Preservation action list** — immediate steps with deadlines (feeds the `legal-hold` workflow)
- **Authentication certificate** — for items ready for judicial use

## Guardrails

- Never alter original evidence files.
- Hash all evidence before any court submission.
- Document every access and transformation.
- Flag any integrity concern immediately.
- Chain-of-custody breaks must be reported, not concealed.
- **Ontology compliance:** outputs must satisfy the litigation-relevant error codes in `core/ontology/error-catalog.md`. If an invariant violation is detected, flag with `[ONT-ERROR: <code>]` and do not produce confident output. I-1 is absolute: no conclusory language.

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.