Findings go in. Verified pull requests come out.
Software Factory takes findings, issues and failing builds through a decision step, a crew of agents and quality gates that are plain rules. People approve the changes that matter. What comes out is a pull request with the run trace and the evidence behind it.
- 01 Findings in. A scanner finding, an issue label, a webhook event, a schedule or a board change starts a run.
- 02 Agents work. 31 built-in agents in 6 crews patch, test and review on a Pix Sandbox, or on a machine you host.
- 03 Gates decide. The quality gate and the release jail are plain rules with no model in the loop.
- 04 People approve. Plans, merge reviews and publish reviews wait for a person. Checkpoints queue everything that needs you, and the cockpit shows the head of the queue.
- 05 Pull request out. A draft pull request arrives with the run trace, the decision rationale and the evidence files.
- 06 Call Pix. Phone Pix to ask how runs are going. Pix reads a plan back before it starts or stops a session, and acts on a plain yes.
Work arrives on the left, passes four stations and leaves as a verified pull request. A person holds the gate at the approval station.
On a call. Pix opens with what is running and what needs you.
After the call. Past calls keep the transcript, and Call Pix picks one up again.
Run it free on your machine · Pix Sandbox from $85 a month · Request a demo
Community includes one self-hosted Belai host. Run belai rc on your machine and the factory runs there.
Start from a hardened base, not an endless patch queue.
Most findings arrive with the base image and the versions you pinned. Safe Harbour moves both to sources with no known vulnerability, exploit or malware report, so those findings never arrive and the weekly patch run gets shorter.
- 01 Compare the base. Each image is matched with hardened bases that run the same runtime, libc and release, scored by Vulnesis and checked against what the vendor claims.
- 02 Get a migration plan. The plan reads the SBOM, the Dockerfile your CI builds and where the image runs, then writes the Dockerfile, pinned packages, registry access and checks.
- 03 Pin clean versions. One policy across every manifest, lockfiles generated for you, and a pull request to review.
Pix finds out why the checks failed on a Safe Harbour pull request, and upgrades a package to its Safe Harbour pick in a repository that uses it.
How Safe Harbour works · Hardened containers in detail · Request a demo
Safe Harbour: Hardened base, same runtime, libc and release, signed and verified.
Governed before it ever reaches your build
The same policy point sits in front of every LLM call your teams make and every dependency they install, so a refusal happens at the proxy rather than in a report written afterwards.
# Pin a blocked version and the answer is a semantic status:
# 423 malware · 426 vulnerable · 425 cooldown · 428 bad actor
$ npm install @example/telemetry-client@2.4.0
npm error code E428
npm error 428 Precondition Required
# Refused before the tarball downloads, so no postinstall
# script ever runs on the build agent.
Add-on · Package Firewall
Prevention, not another finding
A scanner tells you the malicious package is installed. The firewall means it never was. One policy engine sits in front of 24 package registries and the pkg.go.dev API, from npm and PyPI through Debian and RPM to OCI images, Helm charts and Terraform modules. Every refusal names its reason in the status code and the body. The pkg.go.dev API and AUR are free on every plan; every other registry is the Package Firewall add-on.
Set up the firewall · Explore the AI Firewall · What's free · Package Firewall docs · AI Firewall docs
Retire the point tools and their dashboards
The Resolve Platform is an Application Security Posture Management (ASPM) platform that also delivers Unified Vulnerability Management (UVM), Application Security Orchestration and Correlation (ASOC), Continuous Threat Exposure Management (CTEM) and Risk-Based Vulnerability Management (RBVM) in one place, on top of the DevSecOps scanning (SCA, SAST, secrets, IaC and container) you already run.
Application Security Posture Management
One live view of application security posture across every repository, scanner and environment, with an owner and business context attached to each finding.
Unified Vulnerability Management
Every finding from every source deduplicated and normalised into one prioritised queue, instead of a separate backlog per tool.
Application Security Orchestration and Correlation
SARIF, SPDX and CycloneDX from 216 documented security tools, normalised against one schema. Overlapping results from four scanners collapse into a single decision with one owner.
Continuous Threat Exposure Management
Exposure is scoped and scored against reachability and live exploit intelligence, then re-validated as the intelligence moves. Remediation is tracked against your SLA.
Risk-Based Vulnerability Management
Rank open findings against an ordered risk strategy your organisation owns: malware, KEV, EPSS, Coalition ESS, CVSS, CWSS and custom scores.
Drag a rule above another and every open finding re-tiers against it, in the order your organisation decides.
Your existing stack, turned into a live monitoring system
Bring your tools, workflows and data sources onto one platform
Connect the scanners and data sources you already run. Nothing gets ripped out, and nobody has to change how they work.
Apply your business context and GRC frameworks
Set policies that reflect how your environment is actually built and what your auditors ask for.
Take the grind out of the queue
Deduplication, triage and evidence capture run without a human, and Software Factory agents turn findings into pull requests. Checkpoints queue the few things only a person can decide.
See what Software Factory doesTrack performance, decisions and compliance
Board reporting and assessor evidence are byproducts of the daily work. Nothing is reconstructed at quarter end.
See the ten reports it produces31 built-in agents for the work behind a finding
Each agent has one job: scan a repository, patch a finding, re-check a verdict, plan a feature, reproduce a bug, fix a failing test, review a pull request or resolve a merge conflict. Belai, the open source coding harness, runs them, and Pix coordinates them by phone.
- Security fixes: A scout runs the Vulnesis scanners, patchers fix each finding on its own branch, and a verifier re-checks every verdict with the scanner before it is recorded.
- Features, bugs and interface: A planner splits a request into small build items, a designer specifies interface changes as user journeys, a bug investigator names the root cause, and builders and reviewers do the rest.
- Dependencies by ecosystem: 12 agents, each triaging the dependency findings of one group of package ecosystems: Go modules, npm and Deno, Python, Maven and Gradle, Cargo, Composer, Bundler, NuGet, SwiftPM and CocoaPods, CI pipelines, containers and infrastructure files, and the rest.
- Upkeep: Agents that fix failing tests and lint drift, review a named pull request read only, and resolve merge conflicts.
Pix, the AI Firewall and the agent guides in full · Belai, the Apache-2.0 harness behind the agents
Pix CallRun the factory from your phone
Tap the phone icon in the Resolve Platform, or the Call Pix shortcut in the installed app, and talk to Pix. It answers from your factory's own records, and it does nothing to a session until you say yes. The same phone carries every stop the factory has, down to the sandbox container.
- Ask about the factory: What is waiting on you, how runs and sessions are going, which host is offline.
- Start and stop with a read-back: Pix reads back the repository, branch, host and task, and acts on a plain yes. One yes can start up to five sessions.
- Steer a running agent: Send it a message, or answer its permission request or question by voice.
- Spend and savings on request: Spend against your budget, and the same work priced by hand from the rates in your settings, called an estimate.
- Stay on the line: Hold music while it works, and a push notification if a session finishes after you hang up.
What Pix can do on a call · Pix Personal Assistant: the factory from your phone
31 built-in agents in 6 crews
Findings in, verified pull requests out, with people approving.
AI-BOM, AI inventory and governance
See every model, agent and MCP server your teams actually use.
Pipelines, agents, editors, browsers and the phone.
$ vulnetix agent install detects the agents you have and wires each one.
30 CI/CD platforms
GitHub Actions to Tekton, with SARIF out and one exit code. Docs
10 Coding agent add-ons
A dependency guard at the tool call, plus skills and MCP. Docs
13 Editors with the VS Code extension
Findings on the line that causes them, scanned locally. Docs
31 Identifier schemes in the browser
Every CVE on any advisory page, sorted KEV first. Docs
31 Built-in agents in Software Factory
Run them from the Resolve Platform, on a Pix Sandbox or by phone. Docs
Exec-ready clarity
Give CISOs and leadership a live read on AppSec posture without waiting for a slide deck.
Stop re-checking the advisory
Most triage time is spent verifying whether the record is even correct: the affected range, the fix version, whether a patch exists at all. Vulnesis settles that before the ticket opens.
Live GRC visibility
Compliance status is tracked while the work happens. Audit season stops being an archaeology exercise.
Ten reports, produced from the daily work
Nothing here is filled in by hand. Each report is assembled from findings, SBOMs, firewall telemetry and the triage decisions your team already recorded, including the VEX statements Vulnesis mints when someone marks a finding not affected. Period-scoped, org-wide or per repository, exportable as PDF, OSCAL and evidence bundles.
Explore reporting · Read the announcement · Compliance report docs
- EU AI Act: Regulation (EU) 2024/1689
- EU Cyber Resilience Act: Regulation (EU) 2024/2847
- NIST AI RMF: NIST AI RMF 1.0
- ISO 27001: ISO/IEC 27001:2022 (SoA)
- ISO/IEC 42001: ISO/IEC 42001:2023
- OWASP DSOMM: DevSecOps Maturity Model
- OWASP Top 10: OWASP Top 10:2025
- Scan Coverage: Vulnesis Scan Coverage
- PCI DSS: PCI DSS v4.0.1 (ROC)
- Exposure & Remediation: CTEM-aligned
The four sentences we hear on every first call
- 01 · unified visibility: "We can't keep pace with vulnerabilities across multiple tools and environments."
- 02 · contextual prioritisation: "Every tool has its own priority scores, but none account for our business context."
- 03 · manual effort: "We spend more time wrangling data than actually fixing issues."
- 04 · reporting: "Leadership needs clear metrics, but our data is scattered across systems."
Turn vulnerability intelligence into controls the SOC can run today
When a patch is missing, risky, or waiting on a maintenance window, VDB intelligence still gives analysts something concrete to deploy and validate.
Pricing · Browse the malware STIX feeds
- Ingest · STIX 2.1 malware feeds: Per-ecosystem DNS, URL and package indicators, published with sha256 sidecars for SIEM, CTI, EDR and firewall workflows.
- Mitigate · YARA, Suricata and Snort: Deploy detection coverage when no vendor patch or safe package version exists yet.
- Validate · Exploit payload evidence: Use known payloads to confirm controls alert, block or contain the path they claim to cover.
Malware STIX feeds
STIX bundles for CTI platforms, SIEM enrichment, EDR blocklists and network controls.
No-patch detection rules
YARA plus Suricata/Snort rules for vulnerable software that cannot be patched today.
Exploit payload validation
Verify IDS, IPS, WAF, EDR and detection-control effectiveness with mapped payloads.
Malscan on-device checks
Find malicious scripts and known-bad packages already on disk.
Package Firewall coverage
Bad packages refused at download, across 24 registries.
End-of-life visibility
Unsupported runtimes and software surfaced as their own queue.
Built to survive an enterprise estate
- Asset-first design: Every vulnerability arrives already tied to an asset, an owner and its business context.
- The scan does not end with the pipeline: Shipped SBOMs are re-scored daily, so you hear the day an old package turns exploitable.
- Edge architecture: Runs on Cloudflare's edge network, with data residency and latency handled before you ask.
- Same rules as the CLI: Resolve consumes the same OPA rules as the open source CLI. Browse the registry
Slots into the workflows you already run.
Each source connects with one click, and the coverage report shows which are active, stale or missing. Teams keep working the way they do today, so the evidence Vulnesis collects is a side effect of practices already in place rather than a new process to roll out or a business intelligence project to staff.
1 click per source · 216 documented tools · 15 min to first data. See every integration and ecosystem
The platform in action
150+ sources behind one API
Aggregated, normalised and enriched. Query it from the CLI or your coding agent while the decision is still open.
$ vulnetix vdb vuln CVE-2021-44228
Tools we use ourselves
The same binaries and specifications Vulnesis runs in production, ready to inspect, fork or run without an account. CLI · AGPL-3.0, Belai · Apache-2.0, Pix skills · Apache-2.0, SSVC SDKs · MIT, IETF CRIT draft.
Every open source project · What is free · CLI docsIs the vulnerable function ever called?
A tree-sitter call graph from your entry points to the symbol in the advisory, with the working shown either way.
How reachability is computed · Reachability docsEach piece is useful alone. Together they close the loop.
The firewall refuses the malicious package. The CLI proves whether the vulnerable one is reachable. The VDB says whether anyone is exploiting it today. Software Factory agents write and test the fix on a Pix Sandbox and open the pull request. People approve it, and the monitor tells you when the answer changes.
Every step writes a standard file that outlives us: CycloneDX, SPDX, SARIF, OpenVEX, STIX. Take them to a competitor tomorrow if we stop earning the renewal.
Common questions
Which Vulnesis product is right for me?
The Resolve Platform suits enterprise AppSec teams and CISOs who need unified visibility and prioritisation across many tools. Vulnesis SCA is built for engineering teams tracking dependency and package vulnerabilities. The Vulnerability Database (VDB) and free Community tier serve security researchers and tooling teams who need programmatic access to enriched vulnerability intelligence. Software Factory is for teams who want agents to turn findings into verified pull requests that a person approves.
What is Software Factory?
Software Factory is the Vulnesis automation layer. Findings, issues and failing builds go through a decision step, a crew of agents and plain-rule quality gates, and come out as pull requests with their run trace and evidence. People approve the changes that matter. It runs on a Pix Sandbox that Vulnesis operates or on machines you host Belai on, and it does not merge code: a person merges.
Who's the team behind it?
Vulnesis was founded by Christopher Langton (CISO/CTO) and Sean Marshall (CEO). Chris has spent 20+ years in enterprise security and holds a NSW Government Innovation Award. Sean is an exited founder who has been building products for just as long. Before starting Vulnesis, they built internal versions of it for 4 enterprise companies and government organisations.
How is Vulnesis different from other AppSec platforms?
Most platforms unify findings from their own scanners only. The Resolve Platform ingests the tools you already run, and 216 of them are documented one by one, each with a page showing the exact format it emits and how Vulnesis reads it. The open source Vulnesis CLI adds its own SCA, SAST, secrets, IaC, container and malware scanning alongside them. Browse them at /integrations.
How long does it take to set up?
Most teams are up and running within 15 minutes: connect your GitHub org, set a starting policy, and unified vulnerability data begins to flow. There is no collector or sensor to install and nothing for developers to change. A full organisation rollout usually takes a few days.
Ready to make AppSec work the way it should?
The Vulnerability Database, Resolve, SCA and Software Factory cover the whole path from raw advisory to a verified pull request. Each team adopts the piece built for it, on its own timeline.