New from Wiz Research: Critical NGINX flaw. Understand the risk and impact.

CVE-2026-77316: 
Python vulnerability analysis and mitigation

Overview

CVE-2026-77316 is a Server-Side Request Forgery (SSRF) guard bypass vulnerability in Flyto2 Core (PyPI package flyto-core), an execution kernel for automation and AI-agent workflows. The flaw allows an authenticated attacker with control over a hostname's DNS to bypass outbound connection guards via DNS rebinding, reaching internal services or cloud metadata endpoints. All versions up to and including 2.26.7 are affected; the issue is fixed in version 2.26.8. It carries a CVSS v3.1 base score of 7.1 (High) per the GitHub Advisory Database, with the original advisory noting 6.8 (Medium) due to the high attack complexity precondition (Github Advisory, Feedly). The vulnerability was published on July 11, 2026, and added to the GitHub Advisory Database on November 20, 2026 (Github Advisory).

Technical details

The root cause is a Time-of-Check/Time-of-Use (TOCTOU) flaw (CWE-367) combined with CWE-918 (SSRF) in src/core/utils.py. The validate_url_ssrf() function resolves a hostname via socket.getaddrinfo(), checks whether any returned IP is private, and then returns the original hostname string — not the validated IP. Downstream callers (guarded_aiohttp_request(), enforce_outbound_url(), validate_url_with_env_config()) pass this hostname to aiohttp, which performs an independent second DNS resolution at connect time using its own resolver. An attacker who controls an authoritative DNS server can respond with a public IP for the guard's first lookup (passing validation) and then flip the answer to a private/loopback/metadata address (e.g., 127.0.0.1 or 169.254.169.254) for aiohttp's second lookup, bypassing the guard entirely. The attack requires TTL=0 or near-zero DNS behavior and a successful timing-dependent answer flip; static multi-address DNS records are not sufficient because the guard checks all returned IPs (Github Advisory, Security Advisory). A public Python PoC (poc_dns_rebinding.py) is embedded in the advisory and demonstrates the bypass deterministically using a monkeypatched getaddrinfo and a local TCP sentinel (Github Advisory).

Impact

Successful exploitation allows an authenticated attacker to reach internal network services and cloud instance metadata endpoints (e.g., 169.254.169.254) that the SSRF guard is specifically designed to block. The primary impact is high confidentiality loss — an attacker can read responses from internal services, potentially harvesting cloud credentials, internal API tokens, or sensitive configuration data. Integrity impact is rated low (limited write capability via SSRF), and availability is unaffected. The changed scope means the vulnerability can affect resources beyond the Flyto2 Core component itself, including any internal service reachable from the host running the application (Github Advisory, Feedly). Affected modules span a wide surface: http.get/request/batch/paginate, graphql.query/mutation, monitor.http_check, notification.slack/discord/teams, ai.vision_analyze, verify.visual_diff, and browser.proxy_rotate (Security Advisory).

Exploitability

A public proof-of-concept exploit (poc_dns_rebinding.py) is available within the GitHub Security Advisory and demonstrates the bypass against the real guard code (Github Advisory). Feedly classifies the PoC confidence as high, noting it is a runnable Python script that monkeypatches getaddrinfo to simulate DNS rebinding and confirms internal reach (Feedly). There is no evidence of in-the-wild exploitation at this time, no threat actor attribution, and the EPSS score is 0.0 (Feedly). The vulnerability is not listed in the CISA KEV catalog. Exploitation requires low privileges (authenticated access to submit a URL to a guarded module) and attacker-controlled DNS infrastructure with TTL=0 capability, making it moderately complex to execute in practice.

Exploitation steps

  1. Prerequisite — Attacker-controlled DNS: Set up an authoritative DNS server for a domain you control (e.g., rebind.attacker.test) configured with TTL=0. Program the DNS server to return a public IP (e.g., 93.184.216.34) on the first query and a private/internal IP (e.g., 127.0.0.1 or 169.254.169.254) on subsequent queries for the same hostname.
  2. Obtain low-privilege access: Authenticate to the Flyto2 Core instance with any valid low-privilege account that can submit workflow recipes or invoke modules via the MCP execute_module interface.
  3. Trigger a guarded HTTP module: Submit a request to any SSRF-guarded module (e.g., http.get, graphql.query, monitor.http_check, notification.slack) with url=http://rebind.attacker.test:<port>/ pointing to your rebinding hostname.
  4. Guard bypass via DNS flip: The validate_url_ssrf() guard calls socket.getaddrinfo() — your DNS server responds with the public IP. The guard sees a non-private address and allows the request, returning the hostname string.
  5. Internal connection: aiohttp performs its own independent DNS resolution for the actual connection. Your DNS server now responds with the internal target IP (e.g., 169.254.169.254). aiohttp connects to the internal address.
  6. Exfiltrate response: The response from the internal service (e.g., cloud metadata credentials from http://169.254.169.254/latest/meta-data/iam/security-credentials/) is returned to the attacker through the Flyto2 Core module output (Github Advisory, Security Advisory).

Indicators of compromise

  • Network: Outbound HTTP connections from the Flyto2 Core host to 169.254.169.254 (cloud metadata service) or RFC-1918 addresses (10.x.x.x, 172.16-31.x.x, 192.168.x.x) that are not expected in normal operation; DNS queries with TTL=0 responses for external hostnames that subsequently resolve to internal IPs.
  • Logs: Application logs showing validate_url_ssrf ALLOWED for a hostname that later connects to an internal address; repeated requests to guarded modules (http.get, graphql.query, monitor.http_check, etc.) with unusual external hostnames as the target URL; aiohttp connection errors or successful responses from loopback/metadata addresses.
  • Process Behavior: Flyto2 Core process making unexpected TCP connections to loopback (127.0.0.1) or link-local (169.254.x.x) addresses; unusual data returned in module execution results containing cloud credential patterns (e.g., AccessKeyId, SecretAccessKey, Token fields in JSON responses).
  • File System: No specific file artifacts expected from SSRF exploitation alone, but harvested credentials may appear in workflow output logs or recipe execution results stored on disk (Github Advisory).

Mitigation and workarounds

Upgrade flyto-core to version 2.26.8 or later, which is the definitive fix (pip install --upgrade flyto-core) (Release v2.26.8). The fix introduces an _SSRFGuardedResolver that resolves the hostname once, validates all returned IPs at resolve time, and pins the validated address for the actual connection via guarded_client_session() / ssrf_guarded_connector(), eliminating the TOCTOU gap across all affected modules (Patch Commit). As a temporary workaround for environments that cannot immediately upgrade, implement network-level egress controls to block outbound connections from the Flyto2 Core host to RFC-1918 ranges and the link-local metadata address (169.254.169.254), and monitor for unexpected internal service access attempts (Feedly). Operator-approved private network access via FLYTO_ALLOW_PRIVATE_NETWORK and FLYTO_ALLOWED_HOSTS environment variables is preserved in the patched version.

Community reactions

The vulnerability was credited to researcher "Pig-Tail" in the GitHub Security Advisory (Github Advisory). The advisory was noted on infosec.exchange (Mastodon) shortly after publication (infosec.exchange). No significant broader media coverage or vendor statements beyond the official advisory have been identified at this time.

Additional resources


Source: This report was generated using AI

Related Python vulnerabilities:

CVE ID

Severity

Score

Technologies

Component name

CISA KEV exploit

Has fix

Published date

CVE-2026-132275HIGH7.5
  • Python logoPython
  • svglib
NoYesNov 20, 2026
CVE-2026-132276MEDIUM5.9
  • Python logoPython
  • wcmatch
NoYesNov 20, 2026
CVE-2026-132277MEDIUM5.3
  • Python logoPython
  • inscriptis
NoYesNov 20, 2026
CVE-2026-77332LOW2.3
  • Python logoPython
  • memtomem-stm
NoYesNov 20, 2026
CVE-2026-77331LOW2
  • Python logoPython
  • memtomem-stm
NoYesNov 20, 2026

Free Vulnerability Assessment

Benchmark your Cloud Security Posture

Evaluate your cloud security practices across 9 security domains to benchmark your risk level and identify gaps in your defenses.

Request assessment

Additional Wiz resources

Get a personalized demo

Ready to see Wiz in action?

"Best User Experience I have ever seen, provides full visibility to cloud workloads."
David EstlickCISO
"Wiz provides a single pane of glass to see what is going on in our cloud environments."
Adam FletcherChief Security Officer
"We know that if Wiz identifies something as critical, it actually is."
Greg PoniatowskiHead of Threat and Vulnerability Management