Corgea logo

Corgea

Emerging

US DevSecOps AI writing code fixes for Snyk/Semgrep vulnerabilities at 80% faster remediation and 30% less false positives; YC S23 $2.6M Shorooq/Jawed Karim seed Jan 2025 IDC Innovator competing with Snyk and Mobb.ai for AppSec auto-remediation.

41
AI Score
Grade C↑ Trending
AI Visibility Score (Beta)
CybersecurityWebsiteUpdated November 2026

Company Overview

About Corgea

Corgea is a United States-based AI-powered application security automation company — backed by Y Combinator (S23) with $2.6 million in seed funding in January 2025 led by Shorooq Partners with participation from YC, Propeller, Decacorn, Unbound Ventures, Jawed Karim (YouTube co-founder), and Sam Kassoumeh — providing security engineering teams with an automated vulnerability remediation platform that integrates with existing SAST (Static Application Security Testing) tools (Snyk, Semgrep, Checkmarx, SonarQube) to automatically generate AI-written code fixes for identified vulnerabilities, submit pull requests for developer review, and reduce the time from vulnerability detection to remediation by 80% while cutting false positive burden by 30%. Recognized as an IDC Innovator in DevSecOps in November 2024, Corgea serves security teams who face growing vulnerability backlogs that manual remediation cannot clear at the pace of modern software development.

Business Model & Competitive Advantage

Corgea's remediation automation addresses the security engineering bottleneck created by SAST tool proliferation: security-conscious engineering organizations deploy Snyk, Semgrep, or similar SAST scanners that identify hundreds or thousands of potential security vulnerabilities (SQL injection risks, XSS vulnerabilities, insecure deserialization, hardcoded credentials) per scan — but each identified vulnerability requires a human developer to understand the context, write the code fix, test the fix, and submit it through the code review process. Security teams that can't clear vulnerabilities as fast as they're discovered accumulate backlogs where known vulnerabilities age open for months. Corgea's AI automatically analyzes each flagged vulnerability in context (reading the surrounding code, understanding the data flow, identifying the fix pattern appropriate to the vulnerability type and language), generates a syntactically correct code fix, and opens a pull request with the fix and an explanation — enabling developers to review and merge security fixes at 10x the speed of manual remediation.

Competitive Landscape 2025–2026

In 2025, Corgea competes in the DevSecOps automation, vulnerability remediation, and application security platform market with Snyk (application security with limited auto-fix, $530M raised at $7.4B valuation), Veracode (AppSec platform, acquired by Broadcom, NASDAQ: AVGO), and Mobb.ai (AI-powered vulnerability fix, $6M raised) for security engineering team automation adoption. The DevSecOps automation market has grown as organizations face the dual pressure of increasing vulnerability discovery (more developers, more code, more SAST scanning) and decreasing security team capacity relative to the total codebase under management. Jawed Karim's angel investment (YouTube co-founder who has deep engineering infrastructure experience) and Shorooq Partners' MENA-US dual focus reflect both the technical validation and regional expansion potential. Y Combinator S23 backing positions Corgea in the developer tools and security infrastructure investor community. The 2025 strategy focuses on enterprise deployment within existing Snyk and Semgrep customer environments, building the language-specific remediation quality for Java, Python, JavaScript, and Go codebases where vulnerability patterns are most common, and growing the compliance-driven remediation for SOC 2 and PCI DSS security requirement workflows.

Curated content • Fact-checked and verified

Recent Activity

View all →
blog_post
Java4Shell (`CVE-2026-114938`): 64 `A`s and an `=` overflow OpenJDK's AVX-512 Base64 decoder, with pre-auth RCE shown in Keycloak, Jira, Jenkins and Artifactory

OpenJDK's x86-64 Base64 decode intrinsic miscomputes the store mask for a padded tail shorter than four bytes. The output length underflows, BZHI leaves the mask at all ones, and a 64-byte VMOVDQU8 store writes 62 bytes past the decoded array. Calif turned that fixed-value overwrite into a Java heap leak and remote code execution, and reports pre-auth RCE in default Keycloak, Jira, Confluence, Artifactory, Jenkins and PingFederate. It affects OpenJDK 17.0.6 and later and all 21, 25 and 27 builds before the fix, on CPUs with AVX-512 VBMI (Intel Ice Lake and later, AMD Zen 4 and later). Oracle, Corretto and Zulu shipped fixed builds on 17 November. Temurin and Red Hat had not as of 18 November. Until you can patch, -XX:-UseBASE64Intrinsics removes the vulnerable path.

blog_post
Microsoft `agent-framework` 1.26.1 and 1.26.2 on PyPI ran a `.pth` credential stealer that minted Azure, Azure DevOps and GitHub Actions tokens

On 17 November 2026 someone with publishing rights to Microsoft's agent-framework PyPI project uploaded two wheel-only releases, 1.26.1 at 20:35 UTC and 1.26.2 at 21:08 UTC. Each adds a .pth startup file, so the stealer runs every time Python starts in that environment, including pip itself, without anything importing the package. It reads secret-looking environment variables, credential files and other processes' /proc environ, then asks Azure managed identity, Azure DevOps and (in 1.26.2) GitHub Actions for fresh OIDC and access tokens, and posts everything to a webhook.site URL. Both versions are gone from PyPI and 1.26.0 is the last clean release. A PyPI account created the same day, vibiret, is still listed as a maintainer on agent-framework and five sibling projects.

blog_post
needrestart `CVE-2026-124757`: a fake cgroup directory makes a user's process look like root's session, and the notify-send hook runs its name as a root shell command

needrestart, which Debian and Ubuntu run after every apt or dpkg transaction, decides which login session a process belongs to by matching an unanchored regex against /proc/PID/cgroup. A local user can create a directory such as user-0.slice/session-1.scope inside their own delegated systemd cgroup and get filed under root's session. The packaged 400-notify-send hook then builds a su -c shell string from the process's self-chosen name, so on Debian bookworm and Ubuntu 24.04 the next root apt run executes the attacker's commands as root. Upstream fixed it in needrestart 3.11.1 on 17 November 2026. Debian and Ubuntu had no fixed packages on 18 November; Ubuntu rates it High and lists every release from 16.04 to 26.10 as vulnerable.

blog_post
TYPO3 fixes 10 core CVEs: any backend user could re-open the installer and take over the site (`CVE-2026-119623` + `CVE-2026-119622`), and a Form Framework path leak is already exploited

TYPO3 released 14.3.12, 13.4.40, 12.4.52, 11.5.57 and 10.4.63 on 17 November 2026 with ten security advisories. The critical one, CVE-2026-119623 (CVSS 4.0 9.4), let any file whose name started with FIRST_INSTALL re-enable the unauthenticated first-install wizard on a configured site. CVE-2026-119622 gave every backend user a way to create such a file, because the online media endpoint accepted the fallback storage that maps to the public web root without permission checks. Together they let a low-privileged editor create a new administrator and run code. The release also fixes an Import/Export path traversal to RCE, an ImageMagick MVG file read, and CVE-2026-125160, a Form Framework upload-path disclosure TYPO3 says is actively exploited.

blog_post
OPTIMJACK: attackers wrote a card skimmer and crypto clipper into the Optimizely Web snippets of 178 organizations

On 16 November 2026 an attacker abused a flaw in Optimizely's Experimentation account signup flow to add a loader to the Project JavaScript field of customer projects. Optimizely's CDN then served it to visitors of Zoom, the BBC, CNN, Salesforce, GitLab, HSBC, IKEA and at least 171 other organizations for most of a day. The second stage, 1688360[.]cc/track.js, swapped crypto wallet addresses, skimmed card numbers typed into first-party forms, sent cookies and screenshots to the attacker, and routed iPhones and iPads to the Coruna and DarkSword iOS exploit kits.

blog_post
Puppet `CVE-2026-115731` is now in CISA KEV: any agent certificate can run code on the Puppet Server through a crafted report, and open-source Puppet has no fix

CISA added CVE-2026-115731 to its Known Exploited Vulnerabilities catalog on 17 November 2026. The bug, found by Neodyme and fixed on 3 November, is in Puppet's report-processing code: the server rebuilds report events with FromDataConverter, so an `__ptype` value can define a Pcore object type whose RubyMethod annotation is pasted into a class body and run through class_eval. Any host with a CA-signed agent certificate can reach it, and autosign makes that certificate free. The companion bug CVE-2026-115732 lets `evil.target.example.com` authenticate as `target.example.com`. Fixed in Puppet Enterprise 2023.8.13, 2025.12.1 and 2026.0.1, Puppet Core 9.1.3 and 8.22.1, and OpenVox agent 9.3.0/8.32.0 with server 9.2.0/8.18.0. The community `puppet` gem on RubyGems (last release 8.10.0) will not be patched.

blog_post
Arize Phoenix `CVE-2026-53998` (CVSS 10): one unauthenticated GraphQL query reads `os.environ` and `PHOENIX_SECRET` through an f-string template

Arize Phoenix 13.0.0 through 15.6.x rendered user-supplied F_STRING prompt templates with Python's str.format(), so the applyChatTemplate GraphQL query could walk from a template variable through __class__ and __init__.__globals__ to os.environ. The endpoint needs no credentials on default installs. Leaking PHOENIX_SECRET lets an attacker forge admin session tokens and decrypt stored LLM API keys and OAuth tokens. The fix shipped quietly in 15.7.0 on 13 May 2026; the advisory with a working PoC went public on 9 November and the CVE on 16 November.

blog_post
Apache Roller `CVE-2026-129448`: the pre-auth XML-RPC deserialization RCE reaches back to 4.0.0 and is being exploited

On 17 November 2026 Apache published CVE-2026-129448 for Apache Roller 4.0.0 through 6.1.4 and said the issue is being actively exploited. It is the same bug as CVE-2026-82384, disclosed in September for 6.1.5 only: Roller's XML-RPC servlet enabled Apache XML-RPC vendor extensions, so an unauthenticated POST to /roller-services/xmlrpc could carry an ex:serializable Java object that the server deserialized before any authentication or feature check. A public PoC has existed since 28 September. Roller 6.1.6 disables the extensions and closes the endpoint when XML-RPC is off; 6.1.7 is current.

blog_post
Eclipse Mosquitto `CVE-2026-110486`: an empty WebSocket frame turns the 2.1 built-in listener into a pre-auth heap overflow, fixed in 2.1.3 after a public RCE write-up

Mosquitto 2.1.0 through 2.1.2 ship a new built-in WebSocket implementation, enabled by default in CMake builds. When a client sends an empty masked frame, net__read_ws returns the 4 bytes it spent on the masking key as if they were MQTT payload, the packet reader's to_process counter underflows to about 4 GiB, and the next frame is written past the end of the payload buffer. The bug was public as GitHub issue #3704 from 10 August, born0monday published a full pre-auth RCE chain against Home Assistant's add-on on 4 November, and Eclipse shipped the fix on 17 November in 2.1.3 (use 2.1.4). The same release set fixes CVE-2026-126708, where an unchecked hash return accepts any password under OpenSSL 4.0 FIPS, and five lower-severity bugs.

blog_post
Corgea pull request scans are now 5x faster

95% of Corgea pull request scans now finish within 2.0 minutes, down from 10.1 minutes in May, with the full scan and the same accuracy.

blog_post
Weekly Briefing - 17-11-2026

Corgea's weekly briefing for 11-17 November 2026 leads with the WFX Phantom GitHub Actions secret-stealing campaign, then covers Railway's build-platform breach and the OPTIMJACK Optimizely compromise, with grouped links to the rest of the week's newly surfaced research.

blog_post
`@wppconnect-team/wppconnect` `2.4.1` and `@wppconnect/server` `2.10.82` shipped a Telegram secret stealer through trusted publishing, then the git history was rewritten

On 15 and 16 November 2026, unsigned commits under the joaosouz4dev maintainer account added an 'environment diagnostics' module to the wppconnect WhatsApp SDK and its REST server. The projects' own GitHub Actions trusted-publishing workflows shipped @wppconnect-team/wppconnect 2.4.1 (with SLSA provenance) and @wppconnect/server 2.10.82 to npm, plus infected wppconnect/wppconnect-server Docker images. On import or server start the module uploads environment variables, .env and config files, and WhatsApp session tokens to Telegram bot 9034051102. The payload commits were then removed from master and the release tags were moved to look-alike clean commits.

Key Differentiators

Emerging Innovator

Corgea is an emerging player bringing innovative solutions to the Security market.

Frequently Asked Questions

Estimated Visibility Trend (Beta)

Simulated 8-week rolling score

41
↑ Trending

Based on estimated brand signals. Historical tracking coming soon.

Similar Brands

1Password logo

1Password

Security
B2bCybersecuritySaasSecurity

1Password is an enterprise password manager and secrets management platform enabling individuals, teams, and businesses to securely store, manage, and share credentials, credit cards, and sensitive in

Fortinet logo

Fortinet

Security
B2bCybersecuritySaasSecurityPublic

Fortinet is a global leader in network security, providing enterprise firewalls, endpoint protection, network access control, and cloud security products under the FortiGate and Fortinet Security Fabr

Anduril Industries logo

Anduril Industries

Security
B2bCybersecuritySaasSecurityUnicorn

Anduril Industries is a defense technology company building autonomous weapons systems, surveillance infrastructure, and AI-driven defense platforms for the US military and allied nations. Founded in

Splunk logo

Splunk

Security
AnalyticsB2bCybersecurityEnterpriseSaasSecurity

Splunk is a data platform for security and observability founded in 2003 in San Francisco, built on the idea that machine-generated data — logs, events, metrics, traces — contains the intelligence org

Okta logo

Okta

Security
B2bCybersecuritySaasSecurityPublic

Okta is a San Francisco-based identity and access management (IAM) platform — the leading independent identity infrastructure provider — offering workforce identity (employee SSO, MFA, lifecycle manag

SentinelOne logo

SentinelOne

Security
B2bCybersecuritySaasSecurityPublic

SentinelOne is a cybersecurity company providing AI-powered endpoint detection and response (EDR), extended detection and response (XDR), cloud security, and identity security through its Singularity

Compare Corgea with Competitors

Side-by-side AI visibility scores, platform breakdown, and market position.

For Corgea

Claim This Profile

Are you from Corgea? Verify your company email to claim this profile and keep its description, products, and FAQs up to date.

Claim Corgea Profile →
For competitors & analysts

Track AI Visibility in Real Time

Monitor how ChatGPT, Gemini, Perplexity, and Claude mention Corgea vs competitors. Get alerts when AI recommendations shift.

Start Free Tracking →