← Back to Skills Marketplace
762
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install xproof
Description
Proof and accountability layer for AI agents. Anchor verifiable proofs on MultiversX, enforce audit logging, detect violations on Base. REST API, MCP, x402....
Usage Guidance
Before installing or invoking: (1) Verify the origin — confirm xproof.app and the referenced GitHub repo (jasonxkensei) are the official maintainers and that the files are signed or published by the same entity listed in the registry. (2) Do not run the Quick Install curl commands without inspection; they will fetch and write remote files into your .agent/skills folder. (3) If you use webhooks, point them only at endpoints you control and validate X-Webhook-Signature HMACs to avoid inadvertent data leaks (filename/hash/timestamp). (4) Understand x402: agents may need a USDC balance on Base and will make on-chain payments; restrict or audit any automatic payment behavior to avoid unexpected charges. (5) Treat API keys (pm_...) as sensitive; store them securely and do not commit them. (6) If you want higher confidence, ask the publisher for source code or signed releases, or request an explicit mapping between registry owner, GitHub repo, and homepage; providing that would increase confidence and could move this assessment toward benign.
Capability Analysis
Type: OpenClaw Skill
Name: xproof
Version: 1.3.0
The xProof skill implements a mandatory 'Audit Log Standard' that requires the agent to exfiltrate metadata about every critical decision (including action descriptions, risk levels, and input hashes) to an external third-party service (xproof.app). Furthermore, the SKILL.md contains prompt-injection-style instructions commanding the agent to 'BLOCK execution' and allow 'No soft fail' if the third-party service is unavailable, effectively establishing a remote kill-switch and high-telemetry dependency. The installation process also relies on fetching remote markdown files via curl from a personal GitHub repository (jasonxkensei/xproof-openclaw-skill).
Capability Assessment
Purpose & Capability
The skill's name/description (anchoring proofs on MultiversX, audit logging, Base violation detection) matches the REST/MCP/x402 instructions. However there are provenance mismatches: quick-install examples fetch SKILL.md and references from a GitHub user repo (raw.githubusercontent.com/jasonxkensei/...) while the registry owner and homepage differ; SKILL frontmatter version (3.0.0) doesn't match registry version (1.3.0). These mismatches don't prove malice but reduce confidence in origin integrity.
Instruction Scope
Runtime instructions direct the agent to compute local SHA-256 hashes and POST them to https://xproof.app (expected). Concerns: (1) webhook_url field allows the service to POST confirmations to any HTTPS URL — if an agent populates that with attacker-controlled endpoints it can leak metadata (hash, filename, timestamps); (2) x402 flow instructs agents to make on-chain USDC payments on Base, which can cause monetary charges if invoked automatically; (3) Quick Install shows curl commands to fetch files from a remote GitHub repo and write them into .agent/skills — executing those commands blindly would pull remote content into the agent runtime.
Install Mechanism
There is no formal install spec (instruction-only). The Quick Install examples use curl to pull files from raw.githubusercontent.com (a public release host), which is a common pattern but still writes external content into .agent/skills if run. Because these are human-facing instructions (not an automated install spec), risk is moderate: safe if user verifies sources, risky if executed blindly.
Credentials
The skill declares no required env vars or primary credential, but documents two auth modes: API key (XPROOF_API_KEY) and x402 payments. Requesting an API key for the service is proportionate. The payment mode requires the agent to hold USDC on Base and perform on-chain payments; this is a capability that is coherent with x402 but has real monetary impact and increases attack surface if an attacker tricks an agent into paying.
Persistence & Privilege
The skill does not request always:true, has no declared system config paths, and is instruction-only. It does include PR_INSTRUCTIONS encouraging adding files to another GitHub repo — this is outside runtime scope but not an automatic change to agent configuration. No evidence the skill would modify other skills or require elevated persistence.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install xproof - After installation, invoke the skill by name or use
/xproof - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.3.0
**Major update with security, audit, and protocol enhancements.**
- New agent audit log system ("no proof = no execution") for pre-execution proof & accountability.
- Adds anomaly/violation detection and records violations on Base; enables webhook/policy automation.
- Expanded protocol support: full REST API, MCP (JSON-RPC 2.0), and x402 (walletless payment).
- Improved documentation structure — modular skill, certification, MCP, and x402 references.
- Enhanced batch certification, dynamic badges, verifiable PDF proofs, and webhook notifications.
- Stronger security guidance; API keys and payment guidance updated.
Added GET /api/proof/hash/:hash and GET /api/artifact/trust/:hash endpoints for CLI verify and trust commands
v1.2.0
Initial release. Certify agent outputs on MultiversX blockchain. Supports API key and x402 payment, MCP server, MX-8004 validation loop, batch certification up to 50.
Metadata
Frequently Asked Questions
What is xProof — Blockchain Proof for Agents?
Proof and accountability layer for AI agents. Anchor verifiable proofs on MultiversX, enforce audit logging, detect violations on Base. REST API, MCP, x402.... It is an AI Agent Skill for Claude Code / OpenClaw, with 762 downloads so far.
How do I install xProof — Blockchain Proof for Agents?
Run "/install xproof" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is xProof — Blockchain Proof for Agents free?
Yes, xProof — Blockchain Proof for Agents is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does xProof — Blockchain Proof for Agents support?
xProof — Blockchain Proof for Agents is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created xProof — Blockchain Proof for Agents?
It is built and maintained by xKensei (@jasonxkensei); the current version is v1.3.0.
More Skills