Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Orientor

    1

    Enforce a strict zero-decision-authority contract that prevents AI agents from making assumptions without user consent.

    $3

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Orientor

    Orientor

    Example session with this skill installed

    I need to set up a local transcoding pipeline for 4K video. Use the Orientor protocol to start the session.

    • Read your context and instructions
    • Compiled the orientor

    I have analyzed the task and probed your system environment. I've identified several decision points regarding the transcoding stack, codec priorities, and storage limits.

    QUESTIONS.md

    Restatement
    You want to establish a local pipeline for 4K video transcoding. My goal is to configure the tools, scripts, and workflows to handle high-resolution files according to your specific quality and hardware preferences.

    What I found

    • OS: Windows 10 Pro (Build 19045)
    • CPU: AMD Ryzen 9 5900X (12-Core)
    • GPU: NVIDIA GeForce RTX 3080 (10GB VRAM) - NVENC support detected.
    • Tools: ffmpeg version 6.0 (installed), python 3.11.4 (installed), node (missing).
    • Storage: 142GB free on C:, 1.2TB free on D: (NVMe).

    What I will NOT decide without you
    1.

    Hardware Acceleration: Should I prioritize NVENC (GPU) for speed, or libx265 (CPU) for maximum quality/compression ratio?
    2. Container Format: Do you prefer .mp4, .mkv, or .mov as the default output?
    3.

    Workflow Style: Should this be a watched-folder script (auto-transcode on drop) or a CLI-triggered batch utility?
    4. Proxy Generation: Do you need lower-res proxies generated alongside the 4K masters?
    5. Existing Files: If a file already exists in the output path, should I skip, overwrite, or version the filename?

    How to answer
    Edit this file by adding your choices next to each point and reply "done", or simply answer the numbered items directly in the chat.

    Next steps

    1. Review and fill out the QUESTIONS.md file in your root directory.
    2. I will restate the locked plan for final confirmation.
    3. Execution will begin once you approve the plan.
    QUESTIONS.md

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

    About this skill

    Stop fighting your AI agent over unwanted "best practices" and silent configuration choices. This skill enforces a hard zero-decision-authority contract, stripping the agent of its autonomy and returning 100% of the control to the developer. It replaces agent assumptions with a structured, one-time bootstrap protocol and a mandatory "ask-before-acting" rule for every decision point.

    What it does

    • Mandatory Bootstrapping executes a one-time environment probe and research phase to identify every required decision before work starts.
    • Single-Pass Questionnaire generates a dynamic QUESTIONS.md file covering architecture, stack, and risk-weighted choices to avoid mid-task interruptions.
    • Zero-Decision Authority forces the agent to stop and use the TUI question tool for any decision, regardless of how "obvious" or "safe" it seems.
    • Environment Probing measures OS facts, hardware specs, and installed CLI tools automatically instead of asking the user for machine data.
    • Verifiable Execution requires the agent to probe state and verify results before claiming a task is "done" or "fixed."
    • Audit Notifications triggers audible completion alerts for long-running operations like builds, installs, or large downloads.

    How it works

    1. The agent probes your local environment and researches the task to identify gaps in knowledge.
    2. It generates a single QUESTIONS.md file containing a restatement of the goal and a categorized list of every decision needed.
    3. You edit the file or reply in chat to lock in your choices.
    4. The agent executes in verifiable slices, stopping to ask for consent via TUI before any mutation or direction-setting choice.

    Frameworks & tools

    This skill is platform-agnostic and works with any agentic IDE or CLI harness (Claude Code, Cursor, Windsurf). It utilizes local CLI probes, web search for technical research, and standard filesystem operations.

    Why this beats prompting it yourself

    Generic system prompts eventually "drift" as the agent prioritizes its internal training on helpfulness over your specific constraints. This skill uses a formal contract structure and a physical file-based checkpoint (QUESTIONS.md) that anchors the agent's behavior, preventing it from slipping back into making autonomous assumptions.

    Use cases

    • Legacy Refactoring where silent "improvements" to code patterns could break undocumented dependencies.
    • Environment Setup to ensure specific versions of ffmpeg, node, or python are used exactly as intended.
    • Production Hotfixes where the cost of a "safe default" causing a regression is too high.
    • Cloud Infrastructure to prevent accidental creation of expensive resources without explicit confirmation.

    Known limitations

    Requires an environment where the agent has permission to probe system specs and write to the local filesystem. Audible notifications depend on the host machine's audio capabilities.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Prompt & Skill Engineering.

    Requires filesystem write access for QUESTIONS.md. Agent must have permission to run local shell commands for environment probing. Audio alerts require host audio device support.

    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 12 days ago

    Permissions

    Network Access
    Read Files
    Write Files
    Listed12 days ago

    What's inside

    Frequently Asked Questions

    Orientor stops 'hallucinated autonomy' by forcing the agent into a zero-decision-authority contract, replacing silent assumptions with a structured questionnaire and mandatory consent checkpoints.

    This skill is platform-agnostic and designed to work with any agentic environment that has filesystem access, including popular tools like Claude Code, Cursor, and Windsurf.

    The QUESTIONS.md file acts as a physical anchor for the agent's logic; by forcing the agent to document and wait for your input in a file, it overrides the typical 'drift' where agents revert to default training behaviors.

    Orientor requires permissions to execute local CLI probes for environment data (like OS and hardware specs) and the ability to read/write to your project directory to manage the decision logs.

    Yes, while it enforces strict rules, you can define the scope during the initial bootstrap phase, making it ideal for high-stakes tasks like production hotfixes or complex legacy refactoring.

    New in Prompt & Skill Engineering