sentinai guardian

    1

    Zero-trust governance engine for AI agents featuring permission audits, tool call guardrails, and document quarantine.

    $15

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    sentinai guardian

    Example session with this skill installed

    Audit my agent's permissions. It has access to 'Database query' with 'select' and 'drop' actions, and 'CRM export' with no row limits on PII data.

    • Read your context and instructions
    • Compiled the sentinai guardian

    I have audited the agent's permission manifest against the Sentinai policy model.

    Permission Audit Result

    ToolAction/AttributeFindingVerdict
    Database querydropDB-001: Destructive statement blockedDENY
    CRM exportmax_rows: nullDB-020: Bulk exfiltration risk (unbounded PII)HUMAN
    CRM exportpii: trueDB-030: PII projection without explicit limitHUMAN
    GeneralunrecognizedDB-900: Unknown tool or actionHUMAN

    Summary: FAIL
    The audit identified

    1 deny-level finding and

    2 human-escalation findings. The agent is currently over-privileged. You must remove the drop action and define a max_rows limit below 10,000 to meet the automated safety threshold.

    Every finding has been recorded in the acme-core tenant audit chain with a unique decision ID for tracking.

    Next steps

    1. Modify permissions.json to remove the drop permission.
    2. Set a specific max_rows integer for the CRM export tool.
    3. Re-run the audit to confirm the DENY verdict is cleared.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Zero-Trust governance and defence for autonomous AI agents. A guardrail engine that returns allow, deny or human-approval verdicts on tool calls and prompts, with document quarantine scanning, a hash-chained audit log, four-eyes approvals, SEV incident triage, and a permission audit that flags over-broad agent access before the agent runs. Includes the engine, CLI, dashboard template and tests. Experimental / educational — it does not guarantee security or stop any specific attack. The source console stays free on GitHub.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Pattern-based detection only. FNV-1a hash is for demo purposes. Does not guarantee security or stop unknown attack variants. Requires Python 3.10+.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 4 days ago

    Software
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Environment Variables
    Listed4 days ago

    What's inside

    Frequently Asked Questions

    Sentinai Guardian acts as a deterministic security layer that intercepts agent actions, ensuring they can't execute destructive commands or export sensitive data without meeting your specific policy gates.

    This skill is framework-agnostic and built on a zero-dependency Python architecture, featuring specific out-of-the-box integrations for Claude Code and Cursor via command prompts and rule files.

    You receive the full Python-based governance engine, the pattern-matching rule sets, the Four-Eyes approval workflow module, and the tamper-evident logging system.

    Yes, the engine uses a regex-based pattern matching system that allows you to define custom triggers for allow, deny, or human-approval verdicts based on your specific security needs.

    Sentinai Guardian provides a structured, persistent audit trail and deterministic logic that exists outside the agent's context, whereas standard prompts are prone to injection and 'forgetting' constraints.

    Setup involves auditing your agent's permission manifest through the engine and configuring the tool call evaluator to point to your existing agentic workflows.

    New in Agents & Orchestration