NORNR Governance

Put policy before paid actions, require approval for risky autonomous actions, and keep a finance-ready audit trail.

Security Scan

The skill's requirements, instructions, and tiny local bridge align with its stated purpose (governance/approval before paid or risky actions) and delegate real work to a pinned NORNR SDK; nothing in the bundle indicates intentional misdirection.

Purpose & Capability

Name/description (policy/approval/audit for paid actions) match the actual footprint: a lightweight CLI bridge that calls the official NORNR Python SDK (agentpay). Required credential (NORNR_API_KEY) and recommended API scopes (payments:write, approvals:write, audit:read, etc.) are coherent with the declared purpose.

Instruction Scope

SKILL.md only instructs running the local CLI bridge and installing a pinned dependency; it does not ask the agent to read unrelated system files or exfiltrate data. Minor inconsistencies: SKILL.md and README reference optional env vars (NORNR_BASE_URL, NORNR_AGENT_ID) that are not listed in the registry 'required env' metadata, and README contains an accidental local absolute path (non-security-critical but sloppy).

Install Mechanism

No embedded downloads or obscure install hosts: the package is instruction-only and points to a pinned PyPI dependency (nornr-agentpay==0.1.0) via requirements.txt. This is a standard, reviewable install pattern.

Credentials

The single required env (NORNR_API_KEY) and the listed API scopes are consistent with governance functionality, but some scopes (payments:write, webhooks:read, reports:read) are high-privilege. The skill advises creating a dedicated, least-privilege key—which is appropriate and should be followed. Also note the registry metadata only mandates NORNR_API_KEY while documentation references additional optional envs.

Persistence & Privilege

The skill is not force-installed (always:false). It can be invoked autonomously (disable-model-invocation:false), which is the platform default; combined with payments:write this increases blast radius if misconfigured, so enable autonomous invocation only after testing and with restricted API key scope.

Assessment

This bundle appears coherent and delegates governance to the official NORNR SDK, but before installing: (1) review the pinned PyPI package (nornr-agentpay==0.1.0) source to confirm behavior; (2) create a dedicated NORNR API key and restrict scopes to the minimum you need (avoid broad workspace/treasury scopes); (3) test in a non-production workspace and verify that approval-required states actually halt autonomous flows; (4) be cautious about enabling autonomous invocation with a payments-capable key; and (5) note minor documentation issues (optional env vars not listed in registry metadata and an accidental local path in README) — these are not malicious but worth cleaning up.

What this skill is for

Use NORNR as the control layer before an OpenClaw skill triggers a paid action, risky autonomous action, or any downstream step that should leave behind a finance-ready audit trail.

When to use it

• Before a skill triggers a purchase, subscription, or vendor-side paid action
• When an autonomous flow should pause for operator approval before execution
• When finance or operations need one defensible decision record after the action completes
• When suspicious prompt instructions or unusual spend should surface as review posture instead of silent execution

Installation

• python -m pip install -r requirements.txt

Required environment

• NORNR_API_KEY
• NORNR_BASE_URL (optional, defaults to https://nornr.com)
• NORNR_AGENT_ID or a stored NORNR login profile

Recommended API key scope

Minimum action scope for the full skill surface:

• payments:write
• workspace:read
• approvals:write
• events:read
• audit:read

Add these if you want the finance-close paths too:

• reports:read
• webhooks:read

Dependency provenance

This skill delegates governance decisions to the official NORNR Python SDK, agentpay.

• Install source: requirements.txt
• Pinned PyPI package: nornr-agentpay==0.1.0
• Local bridge: nornr_governance.py

Review the pinned SDK revision before use if your environment requires dependency review.

Commands

• python nornr_governance.py preflight --action purchase --amount-usd 25 --counterparty openai --purpose "Run the paid research action"
• python nornr_governance.py approvals
• python nornr_governance.py approve --payment-intent-id pi_123 --comment "Approved after review"
• python nornr_governance.py reject --payment-intent-id pi_123 --comment "Rejected pending review"
• python nornr_governance.py anomalies --counterparty openai
• python nornr_governance.py timeline
• python nornr_governance.py finance-packet
• python nornr_governance.py audit-export
• python nornr_governance.py weekly-review
• python nornr_governance.py monthly-close --provider quickbooks
• python nornr_governance.py review-bundle --counterparty openai

Operating rule

Do not let OpenClaw proceed with the autonomous action until NORNR returns approved or an operator explicitly approves the queued intent. Treat queued, blocked, anomalous, or prompt-risk posture as operator review states, not autonomous green lights.