# Settlement Witness

> OpenClaw skill: fetch replay-stable PASS/FAIL receipts from the public SettlementWitness HTTPS endpoint. Stateless. Deterministic. Signed.

- Skill: `demerzels-lab/settlement-witness` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add demerzels-lab/settlement-witness`
- Raw SKILL.md: https://api.skillmd.com/api/skills/demerzels-lab/settlement-witness/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: Demerzels-lab (https://skillmd.com/u/demerzels-lab)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/demerzels-lab/settlement-witness

---


# SettlementWitness (OpenClaw Skill)

This skill calls the public SettlementWitness HTTPS endpoint. No local services are required.

## What this skill does
When a task is complete and you need proof-of-delivery before settlement, call SettlementWitness to obtain a replay-stable receipt.

SettlementWitness does not judge quality or arbitrate disputes. It only checks whether the provided output matches the provided spec under deterministic rules.

## Canonical endpoints
- POST https://defaultverifier.com/settlement-witness
- GET  https://defaultverifier.com/manifest

## Required input shape
Provide:
- task_id (string)
- spec (object)
- output (object)

## Example request
{
  "task_id": "example-002",
  "spec": { "expected": "foo" },
  "output": { "expected": "foo" }
}

## How to interpret
- If verifier_response.verdict == PASS: verified completion
- If verifier_response.verdict == FAIL: not verified (do not settle automatically)
- receipt_id is the stable identifier to store/log/share

## Safety notes
- Never send secrets (private keys, API keys) in spec/output.
- Keep spec/output minimal and deterministic (hashes/IDs are ideal).

## Install (for OpenClaw users)
Copy this folder into your OpenClaw skills directory as:
settlement-witness/SKILL.md