Skip to content

    Type a page, a command such as keryn scan, or a rule id.

    Partner or OEM

    View as Markdown

    You get: Keryn’s results as standard files your product already reads, the code to consume them, and a first call to the Keryn Cloud API. You need: a current browser; for step 3, a GitHub or Google account to sign in to Keryn Cloud.

    The SDK and thin MCP client are available on npm. See Download for current CLI channel status. The browser workflow exports the sample reports. The shell examples show how to generate and read them locally.

    Download helpdesk.zip, open keryn.app, choose File under Scan and drop the zip. When the scan finishes, open Findings and choose Share and export:

    The Share and export menu of Keryn Web after the helpdesk.zip scan: SARIF 2.1.0, Result JSON, the evidence bundle, Problems for VS Code and the CycloneDX 1.6 AI-BOM, generated in the tab.The Share and export menu of Keryn Web after the helpdesk.zip scan: SARIF 2.1.0, Result JSON, the evidence bundle, Problems for VS Code and the CycloneDX 1.6 AI-BOM, generated in the tab.
    Share and export: SARIF 2.1.0, result JSON, the evidence bundle, VS Code problems and the CycloneDX 1.6 AI-BOM, generated in the tab. Taken from keryn.app main-5g3Rr4q1.js.

    Every file is generated in your tab. SARIF 2.1.0 is what code-scanning products read; CycloneDX 1.6 AI-BOM is the AI inventory; Result JSON is Keryn’s full result, with its schema in the schemas reference.

    The sample’s code

    These are the files in helpdesk.zip, from the AppSec journey.

    helpdesk/tools.py
    import subprocess
    import urllib.request
    def run_command(command: str) -> str:
    """Run a shell command for the agent and return what it printed."""
    done = subprocess.run(command, shell=True, capture_output=True, text=True, timeout=30)
    return done.stdout
    def fetch_page(url: str) -> str:
    """Download a page so the agent can read it."""
    with urllib.request.urlopen(url, timeout=10) as resp:
    return resp.read().decode("utf-8", "replace")
    helpdesk/agent.py
    import os
    from anthropic import Anthropic
    from helpdesk.tools import run_command
    client = Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])
    SYSTEM = "You are a help desk agent. Reply with one shell command that diagnoses the ticket."
    def handle_ticket(ticket_text: str) -> str:
    reply = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=200,
    system=SYSTEM,
    messages=[{"role": "user", "content": ticket_text}],
    )
    command = reply.content[0].text
    return run_command(command)
    anthropic==0.40.0

    Each finding has a stable fingerprint, so you can deduplicate across rescans. These commands read the same formats; the page’s tests run them on files the pre-release CLI writes for the sample, which are the formats Keryn Web exports:

    Terminal window
    keryn scan . -q --sarif keryn.sarif --aibom aibom.cdx.json
    Terminal window
    jq -r '.runs[0].results[] | [.level, .ruleId, (.partialFingerprints | keys[0])] | @tsv' keryn.sarif | sort | uniq -c | sort -rn | head -5
    2 note keryn.inv.lib.anthropic keryn/fingerprint/v1
    1 warning seceng/http_fetch_url keryn/fingerprint/v1
    1 warning seceng.llm05.python-subprocess-shell-true keryn/fingerprint/v1
    1 warning keryn.general.python.urlopen-any-scheme keryn/fingerprint/v1
    1 warning keryn.general.python.subprocess-shell-dynamic-command keryn/fingerprint/v1
    Terminal window
    jq -r '.components[] | "\(.type) \(.name)"' aibom.cdx.json | sort -u
    library anthropic
    machine-learning-model claude-sonnet-4-5

    3. A trust scan on Keryn Cloud (4 minutes)

    Section titled “3. A trust scan on Keryn Cloud (4 minutes)”

    Keryn Cloud runs scans for you on Keryn’s servers. A trust scan of a website reads its public trust pages (security, privacy, sub-processors) and draws no credit.

    1. Open api.keryn.app/account and sign in with GitHub or Google under Other ways to sign in (no email is sent; an email link is also offered). Choose Create organization.
    2. In keryn.app, open Cloud, choose Sign in and approve the code in your account tab.
    3. Choose New scan, Trust site, and enter a site you are allowed to test. The report is stored in your organization.

    The same scan through the API, with an API key from Developers, API keys on the account page (creating one asks for an authenticator code, so set one up under Security first). This needs your own account, so this page’s tests do not run it:

    POST https://api.keryn.app/api/v1/cloud-scans
    Authorization: Bearer keryn_…
    Content-Type: application/json
    {"kind": "trust", "target": "https://your-site.example"}

    The answer is the stored scan with its findings; GET /api/v1/cloud-scans/{id} returns the full report. Every route is in the Cloud API reference. Hosted analysis that spends credit, POST /api/v1/scans, costs 1 cent per page from your starter credit.