OASM AI Studio

An AI security analyst that works inside your attack surface

Ask a question or hand over a task. The agent plans it, runs your scanners, drives a browser, splits the work across subagents — and shows every step while it works. On your infrastructure, with the model you choose.

AI Studio side panel listing three completed subagents with their objectives and run times
Plan complete · 5/5 steps
External exposure review for the CISO
TLS · admin panels · open databases
3 subagents finished
@asset-analyst × 3 — each with its own objective, merged into one report.

Real AI Studio session · fictional demo workspace

14 built-in agentsFrom recon to exploit verification and reports
Built-in skillsBrowser, API testing, Nuclei templates and more
Every tool callLogged with its full input and output
Your modelOpenAI, Anthropic, Gemini or any compatible API

How a task runs

From one sentence to a finished investigation.

You stay in control at every hand-off: the agent proposes, you approve, it runs — and you can stop it at any time.

  1. 01 · ASK

    Describe the goal

    A question, a target to pentest, a template to write — in plain language.

  2. 02 · PLAN

    Approve the plan

    Multi-step tasks start with a written plan you review before it runs.

  3. 03 · RUN

    Watch it work

    Tool calls, subagents and the live browser stream into the side panel.

  4. 04 · REPORT

    Act on the result

    Findings land on real assets and issues, with remediation steps.

AGENT MODE

Hand over the whole task, not just a question.

Switch the composer to Agent and describe the outcome you want. Scope the chat to one project or the whole workspace, pick the model and reasoning effort, or start from a preset.

  • Agent mode plans and runs multi-step work on its own
  • Scope every chat to a project or the whole workspace
  • Choose the model and reasoning effort per conversation
  • Presets: External recon, Ports & services, Web app pentest
AI Studio composer in agent mode, scoped to the Acme Fintech project, with recon presets
PLANS & APPROVALS

Nothing runs until you approve the plan.

Multi-step tasks start with a written plan. Run it automatically, approve each command as it comes, or tell the agent how to change it — and a rule-based tool guard and scope gate check every side-effecting step.

  • Run automatically, approve each command, or reject
  • Edit the plan in plain language before it starts
  • Auto and Manual approval modes per conversation
  • Tool guard and scope gate before anything leaves the sandbox
AI Studio asking to approve a four-step plan before the agent runs it
SUBAGENTS

Big investigations split into parallel workers.

The main agent dispatches background subagents — each with its own objective — and waits for them to report back. Runs keep going on the server even if you close the tab.

  • Each subagent gets a written objective and its own tools
  • Run times and status for every subagent, live
  • A workflow graph shows which plan step each subagent served
  • Stop a run explicitly; closing the tab doesn't
AI Studio workflow view linking plan steps to the subagents that served them, with the shared notebook

Every step in the open

Watch the agent think, not just answer.

A side panel follows each run: the plan, the subagent workflow, every tool call and the cost of each turn.

Plan

The goal and every step, ticked off live as the agent works.

Subagents

Every background subagent with its objective, status and run time.

Tool calls

The full input and output of every call — jump to it in the chat.

Stats

Context used, tokens per second, time to first token and cache hits per turn.

GROUNDED ANSWERS

Answers built from your own evidence.

The agent reads your live inventory and findings, then writes up what it found — grouped by host, ranked by real exploitability, with what it ruled out and where coverage is thin.

  • Summaries tied to real assets and stored evidence
  • Ruled-out items and coverage gaps called out
  • Concrete next steps with owners and deadlines
  • Follow up in the same conversation
AI Studio answer summarising open critical and high findings, grouped by host with stored evidence
PROPOSED FINDINGS

The agent proposes. You decide.

When the agent finds something new, it proposes a finding with the exact request and response as evidence. Proposals wait in Vulnerabilities → Pending review and count for nothing until someone approves them.

  • Every proposal carries replayable evidence
  • AI evidence quotes are checked against stored tool output
  • Approve or reject proposals one by one or in bulk
  • Approved findings join the normal lifecycle and SLAs
Vulnerabilities list filtered to findings proposed by AI and waiting for review
PROJECT CONTEXT

Briefed like a teammate, scoped to the project.

Give each project standing instructions and reference files — scope notes, rules of engagement, API specs. Every chat started from that project follows them and only reads and acts inside it.

  • Instructions the agent follows in every project chat
  • Reference files: scope documents, specs, notes
  • Project chats stay inside the project's assets
  • Start a chat from any project with one click
Project settings with AI agent instructions and reference files

Make it yours

Your agents, your skills, your model.

Everything the agent can do is configurable per workspace — and everything it costs is measured.

Agents

14 built-in agents you can @mention — asset analyst, exploit verifier, penetration testing, report writer — plus your own.

Skills

Skills with built-in methodology: API testing, browser, command execution, exploit verification, Nuclei templates, recon.

Models & providers

OpenAI, Anthropic, Gemini, OpenRouter, DeepSeek, Vercel AI Gateway or any OpenAI-compatible endpoint.

Model defaults

Default, small and subagent models, reasoning effort and the default approval mode.

Connectors

The scanners the agent may run, installed and configured per workspace.

Usage

Tokens, sessions, requests and cache hit rate per day, per model and per conversation.

Live browser

For web-app work the agent drives a headless browser on your worker and shows the page after each command.

Terminal on your workers

Commands run in a locked-down container on a worker — never on the API host.

Web search

Plug in a search provider (Exa, Firecrawl, TinyFish) so the agent can research beyond your data.

Long-term memory

Optionally let the agent remember what matters about your workspace across conversations.

Custom instructions

Workspace-wide instructions added to every conversation's system prompt.

Nuclei template generator

Describe a check in words; the agent drafts the template and saves it to your workspace.

Start anywhere in the console

AI Studio is one click from every page.

TARGETS

Pentest a target

Start an AI pentest straight from a target, with external recon or web-app presets.

TEMPLATES

Write a Nuclei template

Describe the check; the agent drafts the template with OASM AI Studio.

VULNERABILITIES

Explain a finding

Ask AI about any issue for risk, impact and the fix.

SEARCH

Ask instead of filter

Turn a global search into a question about your inventory.

PROJECTS

Brief the agent

Set project instructions and files once; every run uses them.

AI STUDIO

Free-form chat

Ask anything about vulnerabilities, secure coding and best practices.

Self-hosted

Conversations and results stay on your infrastructure.

Project-scoped

The agent only reads and acts inside the project you choose.

Human in the loop

Approve plans, approve each command, or stop a run at any time.

Fully visible

Every tool call, subagent and browser step is shown and kept.

Put an analyst on every finding.

AI Studio ships with every self-hosted OASM. Free, open source, AGPL-3.0.