More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    📦 Package Supply-Chain Sentinel

    1

    Review local dependency manifests for lifecycle-script, source-URL, download-command and version-management clues before installation.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    📦 Package Supply-Chain Sentinel

    📦 Package Supply-Chain Sentinel

    Example session with this skill installed

    Explain how to review the included risky-sample fixture before scanning a real dependency snapshot. Do not install packages or claim a dependency is safe.

    • Read your context and instructions
    • Compiled the package supply-chain sentinel

    Fixture-first review guide

    This is an illustrative workflow explanation, not a captured scan or an assessment of a real dependency.

    1. Ask your connected agent to load SKILL.md, the scanner and the referenced supporting files. For a separately obtained download, extract the complete folder and keep its structure intact.
    2. Use Python 3.10+ to run the scanner against the included risky-sample fixture before providing your own files.
    3. Inspect text-pattern clues for lifecycle-script mentions, non-registry sources, download commands and version-management terms. A match does not establish malicious intent.
    4. Use the supplied checklist for manual package-name, reputation and lockfile-change review. Do not invent download counts or hash-verification results.
    5. Provide only a stable non-secret snapshot. Review output before sharing; filenames may be sensitive.

    Result states: REVIEW means inspect findings. CANNOT_ASSESS means coverage is incomplete. NO_HEURISTIC_FINDINGS is not a safety guarantee.

    Boundary: No package installation, hook execution, registry-hash authentication or Git provenance verification.

    1. Ask your connected agent to load SKILL.md

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

    About this skill

    Review dependency clues before installation

    Package Supply-Chain Sentinel combines a local Python text-pattern scanner with checklists for manual package-name, reputation and lockfile-change review. Automated findings are review clues, not confirmed malicious behavior.

    What you get

    An agent-loaded SKILL.md workflow with a Python scanner, references, fixtures and tests. Your connected host must retrieve the supporting files and execute Python; Agensi is not the Python runtime. Four rules check lifecycle-script mentions, non-registry source references, download commands, and floating-version or package-management clues. It is not a hosted app or a vulnerability database.

    Requirements and version

    Package v1.1.2 requires Python 3.10+ with the standard library and an agent host capable of retrieving supporting files and running Python. A local installation is needed only if your chosen host does not provide that runtime. For a separately obtained download, extract the complete folder. Start with the included fixture. An AI-agent subscription or setup service is not included.

    Reports and limits

    Reports omit source excerpts and matched values. Missing, unreadable or incomplete inputs return CANNOT_ASSESS. Files over 5 MiB are not scanned. Additional matches beyond 500 findings stop scanning with a finding_limit notice. Use only stable, reviewed non-secret snapshots. Filenames can still be sensitive; inspect reports before sharing. This is not a race-resistant sandbox or a total runtime guarantee.

    Separate manual review

    Package-name/typosquatting and lockfile-change checks require manual evidence review. The scanner does not authenticate registry hashes, verify Git provenance, query CVE or reputation services, install packages, execute lifecycle hooks or prove a dependency safe. Absence of findings is not proof of safety or complete coverage.

    Four heuristic text-pattern rules, not a malware verdict or vulnerability database. Package names, reputation and lockfile changes require manual evidence review. No package installation, hook execution, registry-hash authentication, Git provenance verification or safety guarantee. Reports omit source values, but filenames may be sensitive. Incomplete coverage returns CANNOT_ASSESS; no race-resistant sandbox or total runtime guarantee.

    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 26 days ago

    Requires

    Python 3.10+
    Python
    macOS

    Permissions

    Terminal / Shell
    Read Files

    File Scopes

    **/package.json
    **/package-lock.json
    **/pnpm-lock.yaml
    **/yarn.lock
    **/requirements.txt
    **/pyproject.toml
    **/*.lock

    Read-only inspection first. The bundled scanner reads matching manifests and lockfiles and prints markdown or JSON findings. It installs nothing, fetches nothing, transmits nothing, and modifies nothing. Any write, install, deploy, or live-account action requires explicit user confirmation.

    Listed5 months ago
    Updated26 days ago

    What's inside

    Frequently Asked Questions

    This is not a CVE scanner. Four text-pattern rules flag lifecycle-script mentions, source URLs, download commands and version-management clues for review. Package-name, reputation and lockfile-change checks require separate manual evidence. It does not detect intent, authenticate hashes or prove safety.

    The scanner reads supported local text snapshots, including JavaScript and Python dependency manifests and lockfiles. It applies text-pattern rules, not package-manager-specific semantic parsing or registry verification. Review unsupported and incomplete-input warnings.

    You need Python 3.10+, supplied by your agent host or installed locally. Connecting Agensi alone does not supply execution. Start with the included synthetic fixture and review the output before scanning your own non-secret files.

    The approved package includes SKILL.md, a Python scanner, references, fixtures and tests. Your connected agent must retrieve the supporting files needed for the task. A separately purchased download must be extracted with its folder structure intact.

    The scanner returns heuristic findings with input labels, relative paths and line numbers; source excerpts and matched values are omitted. REVIEW means inspect the clues, CANNOT_ASSESS means coverage is incomplete, and NO_HEURISTIC_FINDINGS is not a safety guarantee. Findings do not establish that a line is malicious or prescribe a verified fix.

    New in Security