Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Jev Agent Safety Pro

    1

    Evidence-based security reviews for AI agent commands and code changes.

    $9.90

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Jev Agent Safety Pro

    Jev Agent Safety Pro

    Example session with this skill installed

    Use Jev Agent Safety Pro to review this pull request before merging.

    Environment: CI / Pull Request

    Task: Update the account settings interface.

    The following change was included in the PR:

    diff --git a/src/api/accounts.ts b/src/api/accounts.ts
    @@

    • if (account.ownerId !== user.id) throw new ForbiddenError();
      return serializeAccount(account);

    No regression test results were provided.

    Identify potential security risks, assess the severity, and recommend whether the changes are ready to merge.

    Do not execute or modify any code.

    • Read your context and instructions
    • Compiled the jev agent safety

    JEV AGENT SAFETY PRO — SECURITY REVIEW

    Security Profile: CI / Pull Request
    Execution Mode: AGENT_GUIDED
    Overall Risk: HIGH
    Recommended Decision: REVIEW_REQUIRED
    Runtime Enforcement: None — Advisory Review

    Security Findings

    HIGH — Potential Authorization Regression

    Affected File: src/api/accounts.ts

    Observed Evidence:
    The supplied diff removes an explicit account ownership verification before returning account information.

    Potential Impact:
    If no equivalent authorization check exists elsewhere, an authenticated user could potentially access information belonging to another account.

    Recommended Actions:

    1. Verify authorization enforcement in the endpoint and its callers.
    2. Restore the ownership check if no equivalent protection exists.
    3. Validate access using authorized and unauthorized accounts.
    4. Add regression tests for cross-account access attempts.

    MEDIUM — Missing Regression Test Evidence

    No test execution results were provided.

    Recommended Actions:

    • Verify that authorized users retain access.
    • Test requests involving another user's account.
    • Confirm that unauthorized requests are rejected.
    • Execute the relevant regression tests before merging.

    Review Coverage

    Reviewed:

    • Supplied patch excerpt
    • Account ownership check removal
    • Available task context

    Not Verified:

    • Full pull request
    • Surrounding endpoint implementation
    • Additional authorization middleware
    • Existing test coverage
    • Actual test execution results

    Tests Executed: NOT_RUN

    Final Recommendation

    REVIEW_REQUIRED

    The supplied change presents a potentially significant authorization risk.

    Do not approve the pull request based solely on the available evidence.

    Verify that equivalent authorization protection remains in place and validate the relevant regression tests before merging.

    Security Review Complete — Advisory Assessment

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

    About this skill

    Jev Agent Safety Pro

    Smarter Security Reviews for AI Coding Agents

    AI coding agents can generate powerful code changes, but they can also introduce security risks, unsafe commands, unintended modifications, and potentially dangerous production operations.

    Jev Agent Safety Pro provides structured, evidence-based security review workflows designed to help developers identify risks before accepting AI-generated changes.

    Built around the Jev Developer Harness ecosystem, this native skill brings a consistent security review methodology directly into AI-assisted development.

    What You Get

    🛡️ Command Safety Reviews

    Analyze potentially dangerous shell commands, destructive operations, unexpected network interactions, and production-sensitive actions.

    🔍 Advanced Code & Patch Review

    Review AI-generated changes for authentication vulnerabilities, authorization issues, credential exposure, database risks, and unexpected modifications.

    ⚙️ Environment-Aware Security Profiles

    Apply review procedures tailored to:

    • Local Development
    • CI/CD and Pull Requests
    • Production-Sensitive Environments

    📦 Dependency & Supply Chain Reviews

    Examine package manifests, lockfiles, dependency changes, installation scripts, and potential supply chain risks.

    📊 Evidence-Based Security Reports

    Receive structured assessments containing:

    • Risk severity and explanations
    • Evidence supporting each finding
    • Recommended actions
    • Missing validation requirements
    • Security review recommendations

    🧠 Intelligent Risk Classification

    Distinguish between:

    • PASS_WITH_LIMITATIONS
    • REVIEW_REQUIRED
    • BLOCK_RECOMMENDED
    • INSUFFICIENT_EVIDENCE

    Designed for AI-Assisted Development

    Ideal for developers and teams using AI coding agents to review code changes, evaluate risky operations, and improve development workflows.

    The skill uses the standard SKILL.md format and requires no proprietary scripts, external API keys, or additional runtime dependencies.

    Optional integration with the free jev-dev-harness CLI is supported when available.

    Important Limitations

    Jev Agent Safety Pro Native provides agent-guided security recommendations.

    It does not automatically intercept commands, enforce execution permissions, guarantee vulnerability detection, or replace dedicated security scanners.

    Review quality depends on the available evidence and the capabilities of the connected AI agent.

    No Installation Required

    Jev Agent Safety Pro Native is a fully self-contained AI agent skill.

    ✅ No Node.js installation required. ✅ No proprietary scripts or executables. ✅ No API keys or external accounts required. ✅ No additional MCP servers required. ✅ No Jev Developer Harness installation required.

    Simply activate the skill in a compatible AI coding agent and provide the command, code changes, or Git diff you want reviewed.

    Optional Jev Developer Harness Integration

    Already using the free, open-source jev-dev-harness?

    The skill can optionally incorporate results from its CLI security checks when the CLI is installed, compatible, and accessible to the agent.

    This integration is entirely optional.

    Without Jev, the skill continues to provide its agent-guided security review workflows.

    Note: The Native edition provides advisory security assessments, not automatic command interception or deterministic enforcement.

    Developed by Marcelo Lopes

    Creator and maintainer of the open-source jev-dev-harness project.

    GitHub: https://github.com/celolopes/jev-dev-harness

    Advisory only; no automated execution blocking. Coverage is limited to supplied artifacts and model context. Does not include active vulnerability scanning or deterministic secret detection.

    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 any payment.

    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 27 days ago

    Permissions

    Read Files

    Read-only, agent-guided security reviews. No file writes, network access or external services required.

    Listed27 days ago

    What's inside

    Frequently Asked Questions

    This skill acts as a security gatekeeper for AI agents, using a structured framework to audit shell commands, code patches, and dependency changes for risks like shell injections or credential leaks before they reach your production environment.

    The skill is environment-agnostic and designed to work with any popular AI coding assistant, including Claude Code, Cursor, and Windsurf, by providing a standardized review methodology within the agent's context.

    Yes, while the core framework runs within your agent, you can optionally integrate it with the `jev-dev` CLI to perform local static analysis for deeper verification of code artifacts.

    You will receive the full safety review framework, the risk classification methodology, and documentation on integrating the system into your CI/CD pipelines or local workflows.

    Unlike basic prompting, this skill uses an evidence-led methodology that specifically looks for transitive dependency risks and subtle command redirections that standard LLM instructions often overlook.

    No, the skill is a read-only auditing framework; it provides a PASS, REVIEW, or BLOCK recommendation but does not have the permissions to automatically stop execution or modify your runtime environment.

    New in Security