CVE Watch27 sept 2026

CVE-2026-101058

python-utcp (pip package utcp-http) before 1.1.12 does not verify whether tool URLs declared in a hand-written UTCP manual point at the agen

CVSS

6.9

Medio

EPSS

0.2%

KEV

—

Exploit Today

0-100

Publicado: 27 sept 2026 · Última mod.: 27 sept 2026 · CWE-918

EPSS · 30d

0.2%EPSS · 30 días0.2%

2026-09-282026-10-25

Descripción técnica

python-utcp (pip package utcp-http) before 1.1.12 does not verify whether tool URLs declared in a hand-written UTCP manual point at the agent's own loopback interface when that manual is discovered from a remote, non-loopback origin. Because ensure_secure_url intentionally permits loopback HTTP for local development and native manuals bypassed the loopback check performed by the OpenAPI converter, an attacker who can serve a UTCP manual that a victim registers can cause the client to issue requests to services bound only to 127.0.0.1 on the victim host and have the response bodies returned to the caller (server-side request forgery). The http, sse and streamable_http protocols are all affected. Reach is limited to loopback, and exploitation further requires a loopback service that answers unauthenticated requests with useful data. Fixed in utcp-http 1.1.12, which rejects manuals fetched from a non-loopback origin that declare loopback tool URLs, keyed off the final post-redirect discovery URL.

Referencias oficiales

CVEs relacionados

CVECVSSEPSSKEVRExplotTítuloVis.

CVE-2026-1160696.5 MED

—

——0NiubiGEO before 0.3.2 contains an unauthenticated server-side request forgery vulnerability that allows remote attackers to make the server fetch internal addresses by supplying a crafted domain value to POST /audit-plan. Attackers can submit IP literals, localhost, or private and link-local hosts to read loopback, internal-network, or cloud metadata responses returned in the plan's siteEvidence JSON.13h CVE-2026-1160666.8 MED

—

——0Firecrawl before 2.11.586 contains a server-side request forgery vulnerability in the Playwright service SSRF proxy that allows attackers to bypass isInternalHost checks via DNS rebinding. Attackers controlling DNS for a scrape target can return a public address during validation and a private address at connect time, retrieving cloud metadata and internal resources through the scrape API.13h CVE-2026-1160007.3 ALT

21.2%

——6A security flaw has been discovered in Zylon PrivateGPT 1.0.0/1.0.1. This vulnerability affects the function load_file_from_uri of the file private_gpt/server/ingest/uri_loader.py of the component URI Artifact Loader. Performing a manipulation of the argument Value results in server-side request forgery. It is possible to initiate the attack remotely. The exploit has been released to the public and may be used for attacks. The vendor was contacted early about this disclosure but did not respond in any way.22h CVE-2026-1158866.5 MED

25.2%

——8MLflow is an open source AI engineering platform for agents, large language models, and machine learning models. From 3.12.0rc0 until 3.17.0, the scorer invoke, register, and GenAI evaluate endpoints, including unauthenticated POST /ajax-api/3.0/mlflow/scorer/invoke, accept serialized third-party scorer data that Scorer.model_validate_json resolves using a module-path allowlist without constraining the selected class or its constructor arguments. When optional trulens and trulens-providers-litellm packages are installed, an attacker can select mlflow.genai.scorers.trulens, choose a model URI outside MLflow's native provider registry so execution uses the LiteLLM fallback, and supply api_base and api_key values that cause the server to send an outbound request to an attacker-chosen or internal host with attacker-controlled headers and body. The vulnerable deserialization occurs before a tracking-store lookup or job submission. This issue is fixed in version 3.17.0.2d CVE-2026-1158135.8 MED

29.1%

——9Server-side request forgery in the model download path in Amazon Web Services Multi Model Server 1.1.11 and earlier might allow remote actors to send requests to internal network destinations via a crafted model URL. Multi Model Server is end of life and no fixed version is available. Users should discontinue use of Multi Model Server and migrate to a supported model serving solution.2d CVE-2026-1158107.7 ALT

15.5%

——5Tandoor Recipes before 2.7.0 contains a server-side request forgery vulnerability in the Pestle importer that allows authenticated low-privileged users to force the server to issue GET requests to arbitrary internal or external hosts by crafting a malicious image URL in a Pestle JSON import file. Attackers can exploit the raw requests.get call in cookbook/integration/pestle.py, which bypasses the application's centralized requests-hardened SSRF filter, to reach internal-only endpoints, perform port/service discovery, and exfiltrate internal responses via the stored recipe image.2d