Key advantages

A web security assessment involves preparing application access, exploring features, testing suspected weaknesses, reviewing evidence, and returning to unfinished work. Hunt Web connects these activities through a scanner CLI and a Dashboard, with Hermes Chat for directing further investigation. Its practical value lies in keeping assessment context, execution, findings, and reports associated with identifiable Runs throughout that work.

The capabilities below support different assessment needs:

1. Manage Assessments Across the CLI and Dashboard

Hunt Web began with a command-line workflow and now combines the CLI with a Dashboard that organizes Target → Scope → Run → Findings → Report. The Dashboard adds visible setup, activity tracking, account management, and navigation between assessments and their results.

1. 1TargetAdd the application and choose the testing mode.
2. 2ScopeSave the URLs and account access to test.
3. 3RunPick a profile and follow the scan.
4. 4FindingsReview verified issues and evidence.
5. 5ReportRead the saved snapshot or export a PDF.

This is useful when several assessments need attention. The overview shows active and paused work, recent Runs, and verified findings grouped by severity. Opening a Run leads to its own context and investigation records, so a summary can be followed back to the assessment that produced it.

Enlarge

The installed CLI remains available for terminal-based operations. Use the CLI supplied by the same installation so it receives that installation's configuration and data bindings. CLI-generated results do not automatically create Target/Run registrations in the Dashboard; the shared distribution does not imply automatic import of every separate CLI session.

2. Continue from Recorded Work with Hermes

An initial profile scan can reveal promising features, incomplete comparisons, or candidates that need stronger evidence. Hermes Chat can read the selected Run's previous results and request history, explain its saved state, and accept a focused instruction to continue testing.

For example, a researcher can return to a two-account assessment, inspect the recorded document-sharing requests, and ask Hermes to investigate cross-account access in that workflow. An eligible Run retains its Scope, configured accounts, and earlier findings for the continuation.

Pause/Resume supports returning to saved scanner progress. Hermes adds a conversational way to choose the next investigation, while Chat Templates supplies editable instructions for goals such as authorization, business logic, injection, coverage review, and unfinished candidates.

Enlarge

Reading history, checking current account access, and continuing testing are separate actions. This supports useful context questions while a scan is paused. Execution status and saved outcomes remain available for checking whether an instruction resulted in actual scanner work.

3. Match the Testing Method to Application Access

Hunt Web offers three methods for different starting conditions:

• Blackbox examines the application without supplied authenticated accounts. It suits an assessment of the publicly accessible surface.
• Greybox Environment uses configured account sessions for HTTP requests. It suits authenticated endpoints, APIs, and workflows that can be examined through request/response behavior.
• Greybox Browser uses account-specific interactive Chromium sessions. It suits signed-in pages, navigation, forms, and the traffic produced by application interactions.

The mode determines the access method. Quick, Standard, and Deep determine investigation breadth, budgets, and stopping policy. All three profiles use the same underlying finding verification rules, so selecting a profile changes the investigation plan rather than creating a separate standard for accepting evidence.

Saved Scope records the assessment's application boundaries. A Run retains that Scope, making the tested coverage identifiable when results are reviewed later. Detected application hosts can help prepare subsequent coverage. A Run created with the automatic host policy can also admit related application hosts discovered during that Run, as described in Application hosts and account access; expanding the boundaries beyond that policy, or for a Run without it, requires a new Scope revision and Run.

4. Compare Accounts and Roles Explicitly

Authenticated testing often depends on which identity owns an object or can perform an action. Hunt Web supports one-account and two-account Greybox setup, with separate account sessions or Browsers as appropriate to the mode.

One account provides an authenticated baseline for examining features and permission boundaries. Two accounts support comparisons between owners or roles, including two users with the same privilege level. A document owned by one user, for example, can be compared with an equivalent document owned by another to investigate whether the application enforces ownership.

Account-access checks provide a recorded starting condition. A fresh check through Hermes can establish whether access is still confirmed before further work. This helps distinguish an application behavior from a test that could not proceed because a session was unavailable or expired.

5. Review Evidence Alongside the Finding

Hunt Web keeps candidates, submitted proofs, verification decisions, and verified Findings distinct. Results exposes the available records for a Run, while Findings brings accepted issues together for review.

A finding can include its affected endpoint, impact, technical analysis, reproduction steps, verifier observations, and recommended remediation. These details help a reviewer trace a reported weakness to the evidence behind it and the scan that produced it.

The practical benefit is a reviewable result. An inconclusive candidate can remain visible as unfinished work, while an accepted finding has its own stored identity and verification context. The detailed Findings, Results, and Report guide explains these distinctions and the available evidence views.

6. Carry the Assessment into a Structured Report

The Run's Report provides a saved assessment with its Scope, testing conditions, findings, and coverage limits. Accepted findings from a Hermes continuation can appear in later snapshots of that same Run's report.

PDF provides a formatted document for review, Markdown supports text-based use, and JSON provides structured output. Saved and export timestamps identify the version being read; a report snapshot can lag behind a scan that is still producing results.

The PDF includes a dedicated AI-assisted assessment and human review section. Automated verification provides evidence for review, while a human reviewer still needs to assess reproduction, impact, severity, and remediation in the application's context.

7. Choose the Deployment and Team Access Model

One installer offers Linux Desktop and Linux Server / VPS modes. The tested platform is Ubuntu 24.04 x86_64 with Python 3.12. Desktop supports all three testing methods. Server supports Blackbox and Greybox Environment, with headless Chromium for supported tools; interactive Greybox Browser requires a graphical Desktop session.

This allows the installation to match the work: a Desktop for interactive signed-in exploration, or a Server for supported assessments accessed remotely through a configured deployment. Public VPS access requires domain, HTTPS, and proxy or Tunnel setup. Cloudflare is optional.

Health exposes component readiness and recovery information. Settings and Users & Roles provide workspace configuration and account controls. Admins manage the workspace; Operators use explicitly granted features; Viewers read permitted results. Starting scans, controlling Runs, and sending Hermes instructions are separate permissions, which supports different responsibilities within an assessment team.

Evaluate the Features Against the Intended Assessment

For an evaluation of agentic scanning tools, useful comparison criteria include saved investigation context, authenticated account comparisons, Browser interaction, evidence inspection, reporting, and deployment support. The table above maps Hunt Web's documented capabilities to concrete tasks; it does not rank detection accuracy against another product.

A practical evaluation can follow one application workflow from initial setup through a profile scan, a focused Hermes continuation, and evidence review. That sequence demonstrates how the components work together. Scan outcomes still depend on the configured model, application access, available tools, and testing conditions; feature availability alone does not guarantee complete vulnerability discovery.

For implementation steps, start with the Linux Desktop installation guide or the Linux Server/VPS installation guide, then follow the assessment workflow. The Hermes Chat guide covers directing further investigation from saved Run context.