Agents Dashboard

Overview

Agents gives your team a single, trusted view of the AI work Fencer does on your behalf. As Fencer takes on more of the manual work of security — fixing vulnerabilities, reviewing code, answering questions in Slack, running AI penetration tests — this is where you ask Fencer questions yourself and confirm what each agent did, who asked for it, and what it decided.

It’s designed to give security and engineering teams confidence that AI is making things faster without quietly making decisions you can’t see.

What you get out of it

Confidence that AI work is visible and auditable. Every agent run — successful or not — is recorded with the original request, the agent’s work, and the outcome.
Faster remediation without losing oversight. Engineers can launch a fix with one click, and security leads can follow fixes as they land without chasing tickets.
Triage where your team already works. When someone mentions Fencer in a Slack thread, that conversation is captured alongside the rest of your agent activity.
A clear answer when something goes wrong. Failed runs show their status and error details, with enough context to decide whether to retry or escalate.

Where it fits

Agents lives under Main > Agents in the left-hand navigation. It is part of the Automatic Fixers paid feature and is available to every built-in role. Launching an agent run still respects your existing permissions, so Agents adds visibility without changing who is allowed to act.

Chat with Fencer

Main > Agents opens the chat workspace. Type a question in the Ask Fencer box — for example “what open vulnerabilities do we have on billing-api?” — and press Enter to start a chat. Fencer answers from your organization’s security data: open vulnerabilities, recent security changes, assets, and more. Under Get started with some examples you can pick a ready-made prompt: Summarize security changes, Review open criticals, or Prepare review agenda.

Starting a chat requires the Administrator or Team Member role; Read Only users see You need edit access to start a chat. Chats are private: each one is visible only to the person who started it, and only they can reply in it.

The Agents menu next to the page title lists your chats together with the other agent runs you can see — vulnerability fixes, changeset and codebase security reviews, AI pentests, Slack requests, and more. From the menu you can:

Start a New chat.
Group the list by date, workflow, or status.
Filter to My agents, or by Workflow and Status (Running, Queued, Failed, Completed).
Rename or Delete a chat you own, from the menu next to the chat’s title.

Workflows

Automated vulnerability fixes

Fencer can generate a code-level vulnerability fix and open a pull request on your behalf. From your team’s perspective, this turns a backlog of code findings into a queue of proposed patches your engineers can accept, tweak, or reject — instead of work they have to do from scratch. Each fix runs as an agent session, so View full agent details on a vulnerability opens the fix in the Agents workspace with the vulnerability alongside it.

For the full lifecycle, prerequisites, and supported source-code integrations, see Automated Remediation.

Triage from Slack

When a teammate mentions the Fencer agent in a thread in a connected Slack workspace, the agent can answer questions about your vulnerabilities and update vulnerability priority and asset criticality — without anyone having to switch tools. The full thread is recorded, so security leads keep an audit trail of what was changed and why. Runs from direct messages and private channels are visible only to the person who started them.

To get started, connect Slack via the Slack Integration.

Following along

Open any run from the Agents menu to see the agent’s work and final answer in context. For vulnerability fixes you can also review the proposed code change, download the patch, and open a pull request without leaving the page.

Links from notifications open the Agent activity list, a table of agent runs with the columns Status, Type, Title, Messages, Requested by, and Started. Use its filters to narrow down to failed runs, runs a particular teammate requested, runs in a date range, a title, or a specific type: Slack, Vulnerability Fix, Changeset Security Review, Codebase Security Review, AI Pentest, or Task coordinator.