Audits

An audit is one offensive engagement against a verified app. You choose a mode, review the targets, and Golem takes over.

The modes

Shallow

Baseline scan + asset discovery + live probing. ~45 min.

Deep

Everything in shallow plus the LLM agent’s full attack chain. ~3.5 hrs.

Autonomous

Deep + your own mission brief. ~4 hrs. Plan-gated.

Mobile

Static + dynamic analysis for iOS and Android apps.

How an engagement runs

You launch the audit           →  credits reserved, audit row created
Baseline scanner runs first    →  nuclei templates + security headers + common paths
Asset discovery                →  subdomains enumerated and live-probed
Darkops sandbox provisions     →  hardened environment with the full toolkit
The agent executes (deep)      →  recon → surface map → exploitation → validation
Findings stream in live        →  validated before they're accepted
Finalize                       →  score + coverage summary + Markdown report

You can watch the engagement live from the dashboard, and you get an email (and Slack message, if connected) when the audit completes or fails.

What every audit produces

Findings — confirmed vulnerabilities with severity, evidence, and remediation. See Findings & Evidence.
Security score — a 0–100 score and letter grade, with a full breakdown showing exactly which finding cost how many points. See Security Score.
Coverage summary — every phase and technique attempted, with status (completed / partial / blocked / not applicable) and failures. See below.
Markdown report — executive summary, scope and coverage, findings with evidence, remediation priorities. See Reports.
PDF export — generate a branded PDF of the same report from the dashboard.

Security focus

Security focus areas tell Golem what matters most to you. Pick up to 8 per app (with a per-run override):

Focus	What it directs the agent toward
Data Theft	Injection, broken authorization, IDOR/BOLA, credential and PII exfiltration
Admin Takeover	Auth bypass, privilege escalation, JWT attacks, default creds, exposed admin panels
Data Tampering	Unauthorized record modification, integrity bypass, write-path flaws
Malicious Uploads	Unrestricted file upload leading to RCE, stored XSS, path traversal
Payment Abuse	Price/inventory tampering, coupon abuse, payment workflow bypass
Business Disruption	Denial of service, resource exhaustion
Data Exposure	Public buckets, exposed `.env`/backups, secrets in JS bundles, verbose errors
Integration Abuse	Third-party API misuse, webhook forgery, OAuth/connector privilege issues
Account Takeover	Session fixation, OTP and password-reset flaws, session/token leakage
General Security	Broad coverage across common web and API vulnerability classes

Focus areas are injected into the agent’s prompt for deep audits. Shallow scans don’t consume them. On mobile apps they shape the report emphasis.

Time budgets

Each mode has a fixed task budget; the agent is force-stopped before the wall-clock expires so the report always gets written:

Mode	Task wall-clock	Agent behavior
Shallow	45 min	Baseline + discovery + focused probing
Deep	3.5 hrs	Agent hard-stops at ~92% of budget to write the report
Autonomous	4 hrs	Same hard-stop behavior

Nuclei (post-agent) is capped to the remaining budget so it can never overrun the audit. Even on a forced stop you receive everything confirmed before the cutoff.

Audit statuses

Status	Description
`pending`	Created, task not started yet
`running`	Golem is actively engaging the target
`completed`	Engagement finished, report ready
`partial`	Finished early with incomplete coverage — score is capped at 69
`failed`	Unrecoverable error — confirmed findings are preserved
`cancelled`	Stopped before completion

Authorization

Golem AI assumes every target you submit is authorized for full-scope offensive testing. Apps must pass ownership verification before any audit runs, and launching an audit requires accepting the authorized-testing terms with a signed attestation.

Getting Started

Audits

How It Works

Platform

Billing

API

Overview

Audits

The modes

Shallow

Deep

Autonomous

Mobile

How an engagement runs

What every audit produces

Security focus

Time budgets

Audit statuses

Authorization

Next steps

Shallow Audits

Deep Audits

​Audits

​The modes

Shallow

Deep

Autonomous

Mobile

​How an engagement runs

​What every audit produces

​Security focus

​Time budgets

​Audit statuses

​Authorization

​Next steps

Shallow Audits

Deep Audits

Audits

The modes

How an engagement runs

What every audit produces

Security focus

Time budgets

Audit statuses

Authorization

Next steps