← Back to Skills Marketplace
1779
Downloads
1
Stars
0
Active Installs
11
Versions
Install in OpenClaw
/install settlement-witness
Description
Verify structured agent task outputs with signed receipts and optional TrustScore attribution.
Usage Guidance
This skill is internally coherent and minimal — it simply describes submitting structured task results to a verifier and receiving a signed receipt. Before using it: 1) Verify the external endpoint (https://defaultverifier.com/...) — confirm the operator, TLS certificate, and a privacy/data policy; the SKILL.md's domain looks like a placeholder and may not be trusted by default. 2) Never send secrets or private data; follow the skill's guidance to only submit minimal structured task data and, if possible, send synthetic/test inputs first. 3) Check the public keys URL (/.well-known/sar-keys.json) to confirm signatures can be independently verified and optionally validate a few sample receipts. 4) If you require stronger guarantees, host or run your own verifier implementation rather than relying on an unknown remote service. If you want, provide the actual verifier domain/operator or a sample receipt you received and I can help validate the signature and keys.
Capability Analysis
Type: OpenClaw Skill
Name: settlement-witness
Version: 0.0.11
The 'settlement-witness' skill is designed to provide deterministic verification of agent task outputs via an external API (defaultverifier.com). While it involves sending data to a remote endpoint, the documentation (SKILL.md and README.md) explicitly instructs the agent to only submit minimal structured data and avoid including secrets or private information. There is no evidence of malicious intent, obfuscation, or unauthorized system access.
Capability Assessment
Purpose & Capability
Name and description match the SKILL.md: it describes a deterministic verifier that accepts structured spec/output and returns a verdict and signed receipt. No unrelated binaries, env vars, or install steps are requested.
Instruction Scope
SKILL.md restricts inputs to structured task data and explicitly warns not to include secrets; it does not instruct reading local files or other system state. However, it references submitting data to an external endpoint (https://defaultverifier.com/settlement-witness), which is expected for a remote verifier but worth validating.
Install Mechanism
No install spec and no code files (instruction-only). This minimizes on-disk risk; nothing is downloaded or installed by the skill itself.
Credentials
The skill requests no environment variables, credentials, or config paths. The optional agent identity (wallet address) is consistent with attribution/reputation use described in the doc; no excessive secrets are requested.
Persistence & Privilege
Flags show defaults (not always:true). The skill is user-invocable and can be called autonomously by the agent (normal behavior). It does not request persistent system privileges or modify other skills.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install settlement-witness - After installation, invoke the skill by name or use
/settlement-witness - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.0.11
Improved summary and top-level description for clarity and higher conversion. No functional changes.
v0.0.10
Introduced trust-complete update: added provenance and verifier transparency, clarified structured input model and optional TrustScore attribution, and fixed metadata version mismatch.
v0.0.9
Improved trust layer: clarified that only structured task data is submitted, made TrustScore optional, reduced identity requirements, and added explicit data handling and safety constraints.
v0.0.8
Added data handling section and clarified input structure to address security scan feedback.
v0.0.7
Updated SKILL.md with improved positioning, clearer verification flow, TrustScore integration, and simplified structure.
v0.0.6
v0.0.6
Improved first-run onboarding experience.
Added immediate “Run This Now” execution block
Added concrete success indicators (PASS / receipt_id / sig)
Simplified description to focus on outcome
Reduced friction for first execution
Clarified common failure modes
No protocol or receipt format changes.
v0.0.5
0.0.5 – Usability update
Added Windows PowerShell-safe example (prevents 422 errors).
Added copy/paste REST examples.
Clarified that install ≠ execution (must trigger tools/call or POST).
No API changes.
v0.0.4
0.0.4
- Clarified MCP as preferred integration path
- Added explicit tool name (settlement_witness via tools/call)
- Added spec and SAR public key discovery links
- Documented trailing-slash redirect behavior
- Normalized reason_code example to SPEC_MISMATCH
- Improved determinism wording
v0.0.3
Initial publish: stateless HTTPS witness receipts for Default Settlement Verifier.
v0.0.2
Initial publish: stateless HTTPS witness receipts for Default Settlement Verifier.
v0.0.1
Initial publish: stateless HTTPS witness receipts for Default Settlement Verifier.
Metadata
Frequently Asked Questions
What is SettlementWitness?
Verify structured agent task outputs with signed receipts and optional TrustScore attribution. It is an AI Agent Skill for Claude Code / OpenClaw, with 1779 downloads so far.
How do I install SettlementWitness?
Run "/install settlement-witness" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is SettlementWitness free?
Yes, SettlementWitness is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does SettlementWitness support?
SettlementWitness is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created SettlementWitness?
It is built and maintained by nutstrut (@nutstrut); the current version is v0.0.11.
More Skills