Agent Commerce Action Review

    1

    Turn proposed agent actions into a structured review of approval, target, risk and missing evidence before execution. Instruction-only, with no scripts, network calls or automatic enforcement.

    $5

    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 CLIMuseMuseDotsDots+21 more

    Agent Commerce Action Review

    Example session with this skill installed

    Observed synthetic model-review exercise, October 31, 2026. Fictional request, not an Agensi connector test.

    Review two proposed steps for the Cedar Demo staging site. First, read the public build log for release cedar-17. Second, promote cedar-17 to production at cedar.example.invalid. The user explicitly approved that release and destination in this fictional conversation. Staging tests passed, but nobody has supplied a production rollback procedure or post-release success criteria. Do not perform either step.

    • Read your context and instructions
    • Compiled the agent commerce action

    This is an actual text review of a fictional request, performed October 31, 2026. It is not an Agensi connector result, a blinded evaluation or a full behavioral test suite. No proposed action was executed.

    IDDecisionSeverityPrimary reason
    read-cedar-17-logALLOWlowPublic, read-only inspection of a specified release log.
    promote-cedar-17REVIEWhighProduction promotion lacks a rollback procedure and post-release success criteria.

    read-cedar-17-log: ALLOW

    • Risk categories: read-only.
    • Reasons: The proposal identifies the public build log for Cedar Demo release cedar-17. No sensitive output or external state change is requested. This review does not open the log.
    • Required evidence: None for this advisory classification. If execution is separately authorized, resolve the exact public log location rather than guessing.
    • Execution status: not executed.

    promote-cedar-17: REVIEW

    • Risk categories: production-change, external-publish.
    • Reasons: In this fictional scenario, the stated trusted user approval covers cedar-17 and cedar.example.invalid. Staging tests passed, but that does not establish production success or recoverability. Approval alone does not complete the review.
    • Required evidence: Exact promotion command and immutable release identifier; a rollback procedure and verified prior release; measurable production success criteria, monitoring window and responsible observer. Check whether the operation needs credential access or introduces irreversible changes before any separate execution decision. Do not provide secret values.
    • Execution status: not executed.

    Overall: review remains pending for the promotion. Do not perform either step under this review.

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

    About this skill

    Review an agent's proposed actions before execution. Agent Commerce Action Review checks the stated target, approval evidence, sensitive-data boundaries, validation and recovery plan. It returns an advisory allow, review or deny result for each action, with reasons and the specific evidence still missing.

    Use it for planned deployments, external publishing, payments, destructive changes, billing changes and credential-dependent work. It can review a single proposal or a mixed manifest. A successful review never executes the proposal or grants permission to execute it.

    The complete rubric and examples are included in one instruction file. No code runner, API key or external service is required by the skill. Do not provide secrets, customer records or private wallet material; describe sensitive dependencies with non-secret labels.

    This is Jimmy Holman's instruction-only adaptation of his original MIT-licensed Agent Commerce Guard work, with a standalone review rubric, safety refinements and synthetic examples. It is not the original executable CLI. Advisory review is not technical enforcement or security certification; results depend on supplied facts and model reasoning. Agensi connector compatibility has not yet been exercised.

    Advisory review, not a technical enforcement layer or security certification. Results depend on the facts supplied and the model’s reasoning. The skill cannot prove authorization or inspect real-world state unless that evidence is supplied. It does not replace professional review. Always follow host rules and verify uncertain outcomes before a separate execution decision.

    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 16 days ago

    Permissions

    No special permissions declared or detected

    Listed16 days ago

    What's inside

    Frequently Asked Questions

    No. It produces an advisory review and evidence checklist. It is not a technical enforcement layer.

    No. Execution remains a separate, authorized workflow subject to host rules and current evidence.

    No. Never supply secrets, customer records or private wallet material. Describe sensitive dependencies with non-secret labels.

    No. This version is a complete instruction-only review skill with SKILL.md, README.md and the MIT license. It contains no executable CLI or external service.

    Not yet. Package checks and synthetic text-review exercises have been performed, but actual Agensi connector compatibility has not been exercised.

    New in Agents & Orchestration