Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    shadow agent inventory audit

    1

    Discover unsanctioned AI agents, MCP servers, and LLM scaffolds committed across your git organization.

    $11

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    shadow agent inventory audit

    shadow agent inventory audit

    Example session with this skill installed

    Run an inventory sweep across the 'acme-co' GitHub org. We need to find any LangGraph or MCP configs committed in the last 6 months and see which ones are actually tied to active CI jobs.

    • Read your context and instructions
    • Compiled the shadow agent inventory

    I have completed the inventory sweep for the

    acme-co GitHub organization. I scanned 42 repositories, identifying 7 unique agent configurations across 12 locations.

    Shadow Agent Inventory Report

    FindingRepositoryCategoryConfidence TierLast Modified
    claude_desktop_config.jsonacme-internal-toolsMCP ClientHigh (Active CI)2024-05-12
    agents.yamlmarketing-automationCrewAI ScaffoldHigh (Active CI)2024-06-01
    .mcp.jsondev-sandbox-alphaMCP ClientLow (Config Only)2024-02-15
    langgraph.jsoncustomer-support-botLangGraphHigh (Active CI)2024-05-28
    OAI_CONFIG_LISTresearch-scriptsAutoGenLow (Config Only)2024-03-10

    Coverage Statement
    Scanned 42 of 45 known repos in acme-co's primary GitHub namespace as of 2024-06-07.
    Excluded from this sweep: 3 private repos without access.
    Out of scope by construction: workstation-local or gitignored configs never committed to any repo; agents wired up in interactive sessions; secrets referenced by name (e.g. ${AGENT_ENDPOINT}) where the destination is masked.

    Next steps

    1. Review the High confidence findings in acme-internal-tools and marketing-automation to identify owners.
    2. Run agent-skill-security-scanner on the discovered MCP configs to verify server safety.
    3. Add the verified configurations to your central approved-agent registry.

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

    About this skill

    Engineering teams are losing visibility into the AI agents, MCP servers, and LLM-framework scaffolds their developers are committing to source control. This skill provides a static, file-based inventory sweep to identify unsanctioned AI tooling across an entire GitHub or GitLab organization. It solves the "shadow AI" problem by scanning repositories for specific configurations like .mcp.json, LangGraph scaffolds, and agent-invoking CI jobs that bypass traditional governance.

    What it does

    • Multi-repo scanning identifies agent configurations and framework scaffolds across all accessible organization repositories.
    • Signal verification reads file contents to filter out false positives from documentation examples, test fixtures, or vendored dependencies.
    • Confidence tiering distinguishes between abandoned experimental configs and actively invoked agents corroborated by CI/cron signals.
    • Coverage reporting generates a mandatory audit statement detailing scanned versus total repos and specific blind spots like gitignored local configs.
    • Allowlist cross-referencing marks findings as either matching existing approvals or requiring immediate owner review.

    How it works

    1. Define the sweep scope by enumerating all repositories in the primary GitHub or GitLab namespace and identifying excluded categories like private or archived repos.
    2. Execute pattern matching against the source control for MCP client configs, framework files like agents.yaml, and LLM-calling CI workflows.
    3. Classify findings by opening each hit to verify it is a functional configuration rather than a code comment or tutorial file.
    4. Generate the inventory as a consolidated report or automated script that dedupes findings and assigns confidence tiers.

    Frameworks & tools

    The skill targets MCP (Model Context Protocol), LangGraph, CrewAI, AutoGen, and OpenAI configuration standards. It integrates with GitHub and GitLab APIs for organizational-scale scanning.

    Why this beats prompting it yourself

    Generic prompts often fail to distinguish between a tutorial file and a production agent, leading to high false-positive rates. This skill enforces strict content-reading rules and provides a defensible audit methodology that accounts for repository-scale deduplication and CI-invocation signals.

    Use cases

    • Conduct a pre-audit discovery to see which teams are using unapproved MCP servers.
    • Build a baseline inventory before migrating to an enterprise AI governance platform.
    • Monitor for "config drift" where developers commit new agent scaffolds into existing repos.

    Known limitations

    Cannot detect agents wired in interactive sessions or configs stored in gitignored local directories. Static scanning does not monitor live network traffic or browser extensions.

    Static file scan only. Cannot see interactive sessions, gitignored local configs, or live network traffic. Requires API access to the target git organization repositories.

    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 1 month ago

    Requires

    Supply Chain Gate MCP

    Permissions

    No special permissions declared or detected

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    The skill performs a deep static sweep across your GitHub or GitLab organization to identify unmanaged AI agents, MCP server configurations, and LLM framework scaffolds that bypass traditional security tools. It specifically solves the 'shadow AI' problem by giving security teams a clear inventory of what developers are committing to source control.

    It looks for committed MCP client configs (such as .mcp.json and claude_desktop_config.json), LangGraph, CrewAI and AutoGen scaffolds, and CI or cron jobs that call an agent or an LLM API, using a dated file-signature list. New frameworks appear faster than any list, so the skill treats its patterns as a baseline to re-verify, and its reference carries a review-by date.

    The skill instructions and one dated reference file (the market context and the file-signature table the sweep uses). The free tier is the methodology, a per-repo checklist and a coverage-statement template you can follow by hand. The paid tier has your agent generate a script for your GitHub or GitLab organization that scans the in-scope repositories, merges duplicate findings, and outputs a CSV/JSON inventory with a coverage and blind-spot statement. It can optionally run as a scheduled CI job.

    This skill uses content-level verification to distinguish between functional configurations and non-functional files like documentation or test fixtures. It validates hits against CI/CD workflows and cron signals to ensure you only investigate agents that are actually being invoked.

    No, the skill performs static file-based scanning of your repositories and does not monitor live network traffic, browser extensions, or interactive terminal sessions. It focuses on identifying the underlying scaffolds and persistent configurations committed to your version control.

    Read-only access to the repositories in scope, supplied to the scanning script you run yourself. Use a token limited to read access on the repositories you want covered, and don't paste it into a chat with the agent.

    New in LLM Ops