ucsandmanucsandmanCommunityยท1 Agent Skills Included

DashClaw

Approval gates, policy enforcement, and audit trails for coding agents

Intercepts risky actions from coding agents before they run and routes them to a human for one-click approval. Enforces guard policies on commands like force pushes, database drops, and secret access, blocking dangerous calls fail-closed. Records every decision in a signed, replayable audit ledger with risk scoring, assumption tracking, and compliance exports. Works across Claude Code, Codex, Hermes, MCP, and Node or Python SDKs.
npx skills add ucsandman/DashClaw --all -g -y
Available:

Instructs the agent on the DashClaw governance protocol: start a session, call guard before risky actions, honor block and approval decisions, and record outcomes for the audit ledger.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install DashClaw?โ–ผ

Run `npx skills add ucsandman/DashClaw --all -g -y` in your terminal to install all skills in this suite globally.

How to stop an AI agent from running destructive commands?โ–ผ

DashClaw sits between the agent's decision and tool execution, scoring each action against your policies and blocking or freezing high-risk calls until a human approves them.

Can I approve agent actions remotely?โ–ผ

Yes. Pending actions appear in the Approvals inbox and can be approved or denied from the web dashboard, CLI, a phone PWA, Telegram, or Discord.

Does DashClaw work with Claude Code and Codex?โ–ผ

Yes. It ships PreToolUse hooks for Claude Code, Codex, and Hermes, plus an MCP server and Node and Python SDKs for any other runtime.

Does DashClaw keep an audit trail of agent actions?โ–ผ

Yes. Every decision is written to a signed, replayable ledger with Ed25519 receipts that anyone can verify, and it supports SOC 2, NIST AI RMF, and EU AI Act compliance exports.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’