ArmorCode Zero
Download the ArmorCode Zero Feature Focus to see how Zero triages every finding, groups them into ranked fixes, drafts the pull request with your coding agent, and proves each fix with a rescan.
In this Feature Focus brief, you’ll find:
- How Zero reads every finding and writes the triage call before you open it.
- How findings collapse into fixes ranked by the risk each one removes, with exploited vulnerabilities first.
- How a fix moves from draft pull request to a rescan of the exact commit, with your team approving every change.
A finding is a cost until someone fixes it
Teams got good at finding problems. AI-assisted discovery keeps adding to the backlog, and attackers use AI to compress time to exploit from years to hours. Triage still happens one finding at a time, and “fixed” often means a ticket was closed.
ArmorCode Zero is an AI-native experience built on the ArmorCode platform and the Context Risk Graph. It reads every finding, writes the triage call, and groups findings by the change that resolves them, so one dependency upgrade that clears dozens of findings becomes one fix. Your team directs the work and approves every change. A rescan of the exact commit decides whether a fix counts.
Triage before you open
Every finding arrives with a summary, a recommendation, a confidence level, and a checklist of what to confirm.
Fix ranking
Fixes are ranked by the risk each one removes, with exploited vulnerabilities first.
Proof
Every fix ends confirmed by rescan, saved as a false positive with evidence, or marked blocked.
What are the key capabilities of ArmorCode Zero?
From the triage call to the rescan, plus the speed, choice, and controls around them, these eight capabilities turn a growing backlog into a short list of fixes your team can approve and prove.
- Triage Before You Open the Finding
Every finding carries a summary, a recommendation, a confidence level, and a to-confirm checklist. In internal testing, 99.7% of findings (28,380 of 28,468) arrived pre-triaged, and new or changed findings get theirs in about 30 seconds. - A Short List of Fixes
Findings group by the change that resolves them, within a repo or across repos, so your team triages once per fix. In internal testing, 28.6k findings became about 8.9k fixes, 69% fewer work items. “Act now” surfaces the top three, exploited vulnerabilities first. - Draft Pull Requests Where Engineers Work
Say yes, and Zero drafts the change with Claude Code, Codex, or Muse Code on your Mac or PC, or in the cloud through the GitHub App. It opens a draft pull request for review, and nothing merges without your engineers. One click adds Zero’s local MCP server to those agents, so you can ask your own AI to work with Zero. - Fixes Proven by Rescan
ArmorCode rescans the exact commit, with up to two replacement scans. A fix counts when the scan comes back and the finding is gone. Zero never changes a finding’s status on its own. - Ask in Natural Language
A chat beside every page answers questions and works on live data. Type a request describing the view you want, and Zero builds your custom visuals with drill-downs. - Findings That Open in Under 100 ms
Zero keeps a local cache and loads findings ahead of you, based on where you are headed. All 28.6k findings in the test tenant load cache-first, so you stay in the work instead of waiting on a screen. - Bring Your Own AI
Choose the model for each task from Anthropic, OpenAI, and open-weight providers, or bring your own keys. Keys are tested with the provider and stored encrypted in ArmorCode’s vault, not on your local system. - A Human in Charge
Every consequential change appears as a card you can edit, confirm, or undo. Secrets are redacted before any model call, and local data uses leased-key encryption, where a remote system owns the encryption.
Frequently Asked Questions About ArmorCode Zero
Q: What is ArmorCode Zero?
A: ArmorCode Zero is an AI-native experience built on the ArmorCode platform. It reads every finding, writes the triage call, groups findings into fixes ranked by risk removed, and opens a draft pull request. A rescan confirms each fix. A chat beside every page answers in plain words, and your team approves every change.
Q: What does “fixes, not findings” mean?
A: A finding is a cost until someone fixes it. Zero groups findings by the change that resolves them, so one fix can cover many findings, and ranks each fix by the risk it removes, with exploited vulnerabilities first. Your team triages once per fix instead of once per finding.
Q: How does Zero relate to the Context Risk Graph?
A: Zero is built on the Context Risk Graph, which links findings with assets, code, owners, business services, and controls. That context lets AI reason about your environment, which is how Zero ranks fixes by the risk they remove.
Q: Does Zero change my code or my findings on its own?
A: No. When you say yes, Zero drafts the change and opens a draft pull request for your team to review. Status and triage changes are written back to ArmorCode only with your approval, and Zero follows each write until ArmorCode confirms it.
Q: Which coding tools does Zero work with?
A: Zero can use the coding agent your engineers already run, including Claude Code, Codex, and Muse Code, or run fixes in the cloud through the GitHub App. One click adds Zero’s local MCP server to those agents, so you can ask your own AI to work with Zero. Changes made that way still appear as a card for your approval.
Q: How do I know a fix worked?
A: ArmorCode rescans the exact commit, with up to two replacement scans. The fix counts when a scan comes in and the finding is gone. If a finding is a false positive, Zero saves the evidence. If a fix is blocked, Zero saves a note and offers a Jira ticket.
Q: Does Zero replace the ArmorCode web app or my scanners?
A: No. Zero is another way to work with ArmorCode, and the web app remains the home for the whole organization to govern, administer, and report. Both run on the same platform and the same context. Zero works with your existing ArmorCode tenant, GitHub, Jira, and the scanners you already run across 425+ integrations. It does not scan your code itself.
Q: Which AI models does Zero use, and how is my data handled?
A: Zero uses models from Anthropic, OpenAI, and open-weight providers, with a model chosen for each task. You can bring your own model keys, which are tested with the provider and stored encrypted in ArmorCode’s vault, not on your local system. Secrets are redacted before any model call. Zero keeps a local cache so findings open in under 100 ms, using leased-key encryption, and your ArmorCode tenant remains the system of record.
Q: How do I get access?
A: ArmorCode Zero is in limited availability to select customers through our Design Partner Program and launches as a desktop app for Mac or PC. Join the waitlist or talk to your ArmorCode account team.
Key Takeaways
- ArmorCode Zero is an AI-native experience built on the ArmorCode platform and the Context Risk Graph.
- Every finding arrives triaged, with a summary, recommendation, confidence level, and to-confirm checklist.
- Findings group into fixes ranked by the risk each one removes, so teams triage once per fix.
- Zero drafts the change with Claude Code, Codex, Muse Code, or in the cloud, and opens a draft pull request for review.
- A fix counts only when ArmorCode rescans the exact commit, and your team approves every change.
