CVE Feed

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the gateway's /api/approval/allow-list endpoint permits unauthenticated modification of the tool approval allowlist when no auth_token is configured (the default). By adding dangerous tool names (e.g., shell_exec, file_write) to the allowlist, an attacker can cause the ExecApprovalManager to auto-approve all future agent invocations of those tools, bypassing the human-in-the-loop safety mechanism that the approval system is specifically designed to enforce. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the _safe_extractall() function in PraisonAI's recipe registry validates archive members against path traversal attacks but performs no checks on individual member sizes, cumulative extracted size, or member count before calling tar.extractall(). An attacker can publish a malicious recipe bundle containing highly compressible data (e.g., 10GB of zeros compressing to ~10MB) that exhausts the victim's disk when pulled via LocalRegistry.pull() or HttpRegistry.pull(). This vulnerability is fixed in 4.5.128.

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, read_skill_file() in skill_tools.py allows reading arbitrary files from the filesystem by accepting an unrestricted skill_path parameter. Unlike file_tools.read_file which enforces workspace boundary confinement, and unlike run_skill_script which requires critical-level approval, read_skill_file has neither protection. An agent influenced by prompt injection can exfiltrate sensitive files without triggering any approval prompt. This vulnerability is fixed in 1.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the /media-stream WebSocket endpoint in PraisonAI's call module accepts connections from any client without authentication or Twilio signature validation. Each connection opens an authenticated session to OpenAI's Realtime API using the server's API key. There are no limits on concurrent connections, message rate, or message size, allowing an unauthenticated attacker to exhaust server resources and drain the victim's OpenAI API credits. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the WSGI-based recipe registry server (server.py) reads the entire HTTP request body into memory based on the client-supplied Content-Length header with no upper bound. Combined with authentication being disabled by default (no token configured), any local process can send arbitrarily large POST requests to exhaust server memory and cause a denial of service. The Starlette-based server (serve.py) has RequestSizeLimitMiddleware with a 10MB limit, but the WSGI server lacks any equivalent protection. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the /api/v1/runs endpoint accepts an arbitrary webhook_url in the request body with no URL validation. When a submitted job completes (success or failure), the server makes an HTTP POST request to this URL using httpx.AsyncClient. An unauthenticated attacker can use this to make the server send POST requests to arbitrary internal or external destinations, enabling SSRF against cloud metadata services, internal APIs, and other network-adjacent services. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, deploy.py constructs a single comma-delimited string for the gcloud run deploy --set-env-vars argument by directly interpolating openai_model, openai_key, and openai_base without validating that these values do not contain commas. gcloud uses a comma as the key-value pair separator for --set-env-vars. A comma in any of the three values causes gcloud to parse the trailing text as additional KEY=VALUE definitions, injecting arbitrary environment variables into the deployed Cloud Run service. This vulnerability is fixed in 4.5.128.

PraisonAI is a multi-agent teams system. Prior to 4.5.128, the Flask API endpoint in src/praisonai/api.py renders agent output as HTML without effective sanitization. The _sanitize_html function relies on the nh3 library, which is not listed as a required or optional dependency in pyproject.toml. When nh3 is absent (the default installation), the sanitizer is a no-op that returns HTML unchanged. An attacker who can influence agent input (via RAG data poisoning, web scraping results, or prompt injection) can inject arbitrary JavaScript that executes in the browser of anyone viewing the API output. This vulnerability is fixed in 4.5.128.

URI nameConstraints from constrained intermediate CAs are parsed but not enforced during certificate chain verification in wolfcrypt/src/asn.c. A compromised or malicious sub-CA could issue leaf certificates with URI SAN entries that violate the nameConstraints of the issuing CA, and wolfSSL would accept them as valid.

A vulnerability has been found in D-Link DIR-605L 2.13B01. This affects the function formAdvFirewall of the file /goform/formAdvFirewall of the component POST Request Handler. Such manipulation of the argument curTime leads to buffer overflow. The attack may be launched remotely. The exploit has been disclosed to the public and may be used. This vulnerability only affects products that are no longer supported by the maintainer.

PraisonAIAgents is a multi-agent teams system. Prior to 1.5.128, he memory hooks executor in praisonaiagents passes a user-controlled command string directly to subprocess.run() with shell=True at src/praisonai-agents/praisonaiagents/memory/hooks.py. No sanitization is performed and shell metacharacters are interpreted by /bin/sh before the intended command executes. Two independent attack surfaces exist. The first is via pre_run_command and post_run_command hook event types registered through the hooks configuration. The second and more severe surface is the .praisonai/hooks.json lifecycle configuration, where hooks registered for events such as BEFORE_TOOL and AFTER_TOOL fire automatically during agent operation. An agent that gains file-write access through prompt injection can overwrite .praisonai/hooks.json and have its payload execute silently at every subsequent lifecycle event without further user interaction. This vulnerability is fixed in 1.5.128.

Flux notification-controller is the event forwarder and notification dispatcher for the GitOps Toolkit controllers. Prior to 1.8.3, the gcr Receiver type in Flux notification-controller does not validate the email claim of Google OIDC tokens used for Pub/Sub push authentication. This allows any valid Google-issued token, to authenticate against the Receiver webhook endpoint, triggering unauthorized Flux reconciliations. Exploitation requires the attacker to know the Receiver's webhook URL. The webhook path is generated as /hook/sha256sum(token+name+namespace), where the token is a random string stored in a Kubernetes Secret. There is no API or endpoint that enumerates webhook URLs. An attacker cannot discover the path without either having access to the cluster and permissions to read the Receiver's .status.webhookPath in the target namespace, or obtaining the URL through other means (e.g. leaked secrets or access to Pub/Sub config). Upon successful authentication, the controller triggers a reconciliation for all resources listed in the Receiver's .spec.resources. However, the practical impact is limited: Flux reconciliation is idempotent, so if the desired state in the configured sources (Git, OCI, Helm) has not changed, the reconciliation results in a no-op with no effect on cluster state. Additionally, Flux controllers deduplicate reconciliation requests, sending many requests in a short period results in only a single reconciliation being processed. This vulnerability is fixed in 1.8.3.

SiYuan is a personal knowledge management system. Prior to 3.6.4, SiYuan configures Mermaid.js with securityLevel: "loose" and htmlLabels: true. In this mode, <img> tags with src attributes survive Mermaid's internal DOMPurify and land in SVG <foreignObject> blocks. The SVG is injected via innerHTML with no secondary sanitization. When a victim opens a note containing a malicious Mermaid diagram, the Electron client fetches the URL. On Windows, a protocol-relative URL (//attacker.com/image.png) resolves as a UNC path (\\attacker.com\image.png). Windows attempts SMB authentication automatically, sending the victim's NTLMv2 hash to the attacker. This vulnerability is fixed in 3.6.4.

In wolfSSL, ARIA-GCM cipher suites used in TLS 1.2 and DTLS 1.2 reuse an identical 12-byte GCM nonce for every application-data record. Because wc_AriaEncrypt is stateless and passes the caller-supplied IV verbatim to the MagicCrypto SDK with no internal counter, and because the explicit IV is zero-initialized at session setup and never incremented in non-FIPS builds. This vulnerability affects wolfSSL builds configured with --enable-aria and the proprietary MagicCrypto SDK (a non-default, opt-in configuration required for Korean regulatory deployments). AES-GCM is not affected because wc_AesGcmEncrypt_ex maintains an internal invocation counter independently of the call-site guard.

Helm is a package manager for Charts for Kubernetes. In Helm versions <=3.20.1 and <=4.1.3, a specially crafted Chart will cause helm pull --untar [chart URL | repo/chartname] to write the Chart's contents to the immediate output directory (as defaulted to the current working directory; or as given by the --destination and --untardir flags), rather than the expected output directory suffixed by the chart's name. This vulnerability is fixed in 3.20.2 and 4.1.4.

A flaw has been found in D-Link DIR-605L 2.13B01. Affected by this issue is the function formSetMACFilter of the file /goform/formSetMACFilter of the component POST Request Handler. This manipulation of the argument curTime causes buffer overflow. The attack may be initiated remotely. The exploit has been published and may be used. This vulnerability only affects products that are no longer supported by the maintainer.

Joomla HikaShop 4.7.4 contains a reflected cross-site scripting vulnerability that allows unauthenticated attackers to inject malicious scripts by manipulating GET parameters in the product filter endpoint. Attackers can craft malicious URLs containing XSS payloads in the from_option, from_ctrl, from_task, or from_itemid parameters to steal session tokens or login credentials when victims visit the link.

Joomla Solidres 2.13.3 contains a reflected cross-site scripting vulnerability that allows unauthenticated attackers to inject malicious scripts by manipulating multiple GET parameters including show, reviews, type_id, distance, facilities, categories, prices, location, and Itemid. Attackers can craft malicious URLs containing JavaScript payloads in these parameters to steal session tokens, login credentials, or manipulate site content when victims visit the crafted links.

Joomla VirtueMart Shopping-Cart 4.0.12 contains a reflected cross-site scripting vulnerability that allows attackers to inject malicious scripts by manipulating the keyword parameter. Attackers can craft malicious URLs containing script payloads in the keyword parameter of the product-variants endpoint to execute arbitrary JavaScript in victim browsers and steal session tokens or credentials.

Joomla iProperty Real Estate 4.1.1 contains a reflected cross-site scripting vulnerability that allows attackers to inject malicious scripts by manipulating the filter_keyword parameter. Attackers can craft URLs containing JavaScript payloads in the filter_keyword GET parameter of the all-properties-with-map endpoint to execute arbitrary code in victim browsers and steal session tokens or credentials.

Joomla JLex Review 6.0.1 contains a reflected cross-site scripting vulnerability that allows attackers to inject malicious scripts by manipulating the review_id URL parameter. Attackers can craft malicious links containing JavaScript payloads that execute in victims' browsers when clicked, enabling session hijacking or credential theft.

WordPress adivaha Travel Plugin 2.3 contains a time-based blind SQL injection vulnerability that allows unauthenticated attackers to manipulate database queries by injecting SQL code through the 'pid' GET parameter. Attackers can send requests to the /mobile-app/v3/ endpoint with crafted 'pid' values using XOR-based payloads to extract sensitive database information or cause denial of service.

WordPress adivaha Travel Plugin 2.3 contains a reflected cross-site scripting vulnerability that allows unauthenticated attackers to inject malicious scripts by manipulating the isMobile parameter. Attackers can craft malicious URLs containing JavaScript payloads in the isMobile GET parameter at the /mobile-app/v3/ endpoint to execute arbitrary code in victims' browsers and steal session tokens or credentials.

A vulnerability was detected in D-Link DIR-605L 2.13B01. Affected by this vulnerability is the function formVirtualServ of the file /goform/formVirtualServ of the component POST Request Handler. The manipulation of the argument curTime results in buffer overflow. The attack can be launched remotely. The exploit is now public and may be used. This vulnerability only affects products that are no longer supported by the maintainer.

A security vulnerability has been detected in Totolink A7100RU 7.4cu.2313_b20191024. Affected is the function setWiFiAclRules of the file /cgi-bin/cstecgi.cgi of the component CGI Handler. The manipulation of the argument mode leads to os command injection. The attack can be initiated remotely. The exploit has been disclosed publicly and may be used.

nimiq-blockchain provides persistent block storage for Nimiq's Rust implementation. In 1.3.0 and earlier, block timestamp validation enforces that timestamp >= parent.timestamp for non-skip blocks and timestamp == parent.timestamp + MIN_PRODUCER_TIMEOUT for skip blocks, but there is no visible upper bound check against the wall clock. A malicious block-producing validator can set block timestamps arbitrarily far in the future. This directly affects reward calculations via Policy::supply_at() and batch_delay() in blockchain/src/reward.rs, inflating the monetary supply beyond the intended emission schedule.

A weakness has been identified in Totolink A7100RU 7.4cu.2313_b20191024. This impacts the function setWiFiBasicCfg of the file /cgi-bin/cstecgi.cgi of the component CGI Handler. Executing a manipulation of the argument wifiOff can lead to os command injection. It is possible to launch the attack remotely. The exploit has been made available to the public and could be used for attacks.

Heap buffer overflow in CertFromX509 via AuthorityKeyIdentifier size confusion. A heap buffer overflow occurs when converting an X.509 certificate internally due to incorrect size handling of the AuthorityKeyIdentifier extension.

A low-privileged remote attacker can send Modbus packets to manipulate register values that are inputs to the odorant injection logic such that too much or too little odorant is injected into a gas line.

A security flaw has been discovered in Totolink A7100RU 7.4cu.2313_b20191024. This affects the function setStorageCfg of the file /cgi-bin/cstecgi.cgi of the component CGI Handler. Performing a manipulation of the argument sambaEnabled results in os command injection. It is possible to initiate the attack remotely. The exploit has been released to the public and may be used for attacks.

An attacker could use data obtained by sniffing the network traffic to forge packets in order to make arbitrary requests to Contemporary Controls BASC 20T.

Two potential heap out-of-bounds write locations existed in DecodeObjectId() in wolfcrypt/src/asn.c. First, a bounds check only validates one available slot before writing two OID arc values (out[0] and out[1]), enabling a 2-byte out-of-bounds write when outSz equals 1. Second, multiple callers pass sizeof(decOid) (64 bytes on 64-bit platforms) instead of the element count MAX_OID_SZ (32), causing the function to accept crafted OIDs with 33 or more arcs that write past the end of the allocated buffer.

A vulnerability was identified in Totolink A7100RU 7.4cu.2313_b20191024. The impacted element is the function setDmzCfg of the file /cgi-bin/cstecgi.cgi of the component CGI Handler. Such manipulation of the argument wanIdx leads to os command injection. The attack may be performed from remote. The exploit is publicly available and might be used.

PraisonAI is a multi-agent teams system. Prior to 4.5.121, the execute_command function and workflow shell execution are exposed to user-controlled input via agent workflows, YAML definitions, and LLM-generated tool calls, allowing attackers to inject arbitrary shell commands through shell metacharacters. This vulnerability is fixed in 4.5.121.

Sonicverse is a Self-hosted Docker Compose stack for live radio streaming. The Sonicverse Radio Audio Streaming Stack dashboard contains a Server-Side Request Forgery (SSRF) vulnerability in its API client (apps/dashboard/lib/api.ts). Installations created using the provided install.sh script (including the one‑liner bash <(curl -fsSL https://sonicverse.short.gy/install-audiostack)) are affected. In these deployments, the dashboard accepts user-controlled URLs and passes them directly to a server-side HTTP client without sufficient validation. An authenticated operator can abuse this to make arbitrary HTTP requests from the dashboard backend to internal or external systems. This vulnerability is fixed with commit cb1ddbacafcb441549fe87d3eeabdb6a085325e4.

Apollo MCP Server is a Model Context Protocol server that exposes GraphQL operations as MCP tools. Prior to version 1.7.0, the Apollo MCP Server did not validate the Host header on incoming HTTP requests when using StreamableHTTP transport. In configurations where an HTTP-based MCP server is run on localhost without additional authentication or network-level controls, this could potentially allow a malicious website—visited by a user running the server locally—to use DNS rebinding techniques to bypass same-origin policy restrictions and issue requests to the local MCP server. If successfully exploited, this could allow an attacker to invoke tools or access resources exposed by the MCP server on behalf of the local user. This issue is limited to HTTP-based transport modes (StreamableHTTP). It does not affect servers using stdio transport. The practical risk is further reduced in deployments that use authentication, network-level access controls, or are not bound to localhost. This vulnerability is fixed in 1.7.0.

CLIENT_CERT authentication does not fail as expected for some scenarios when soft fail is disabled and FFM is used in Apache Tomcat. This issue affects Apache Tomcat: from 11.0.0-M14 through 11.0.20, from 10.1.22 through 10.1.53, from 9.0.92 through 9.0.116. Users are recommended to upgrade to version 11.0.21, 10.1.54 or 9.0.117, which fixes the issue.

Insertion of Sensitive Information into Log File vulnerability in the cloud membership for clustering component of Apache Tomcat exposed the Kubernetes bearer token. This issue affects Apache Tomcat: from 11.0.0-M1 through 11.0.20, from 10.1.0-M1 through 10.1.53, from 9.0.13 through 9.0.116. Users are recommended to upgrade to version 11.0.21, 10.1.54 or 9.0.117, which fix the issue.

Missing Encryption of Sensitive Data vulnerability in Apache Tomcat due to the fix for CVE-2026-29146 allowing the bypass of the EncryptInterceptor. This issue affects Apache Tomcat: 11.0.20, 10.1.53, 9.0.116. Users are recommended to upgrade to version 11.0.21, 10.1.54 or 9.0.117, which fix the issue.

LangChain is a framework for building agents and LLM-powered applications. Prior to 0.3.84 and 1.2.28, LangChain's f-string prompt-template validation was incomplete in two respects. First, some prompt template classes accepted f-string templates and formatted them without enforcing the same attribute-access validation as PromptTemplate. In particular, DictPromptTemplate and ImagePromptTemplate could accept templates containing attribute access or indexing expressions and subsequently evaluate those expressions during formatting. Second, f-string validation based on parsed top-level field names did not reject nested replacement fields inside format specifiers. In this pattern, the nested replacement field appears in the format specifier rather than in the top-level field name. As a result, earlier validation based on parsed field names did not reject the template even though Python formatting would still attempt to resolve the nested expression at runtime. This vulnerability is fixed in 0.3.84 and 1.2.28.

Improper Encoding or Escaping of Output vulnerability in the JsonAccessLogValve component of Apache Tomcat. This issue affects Apache Tomcat: from 11.0.0-M1 through 11.0.20, from 10.1.0-M1 through 10.1.53, from 9.0.40 through 9.0.116. Users are recommended to upgrade to version 11.0.21, 10.1.54 or 9.0.117 , which fix the issue.

Missing hash/digest size and OID checks allow digests smaller than allowed when verifying ECDSA certificates, or smaller than is appropriate for the relevant key type, to be accepted by signature verification functions. This could lead to reduced security of ECDSA certificate-based authentication if the public CA key used is also known. This affects ECDSA/ECC verification when EdDSA or ML-DSA is also enabled.

A vulnerability was determined in FoundationAgents MetaGPT up to 0.8.1. The affected element is the function Bash.run in the library metagpt/tools/libs/terminal.py. This manipulation causes os command injection. The attack is possible to be carried out remotely. The project was informed of the problem early through a pull request but has not reacted yet.

Beszel is a server monitoring platform. Prior to 0.18.7, some API endpoints in the Beszel hub accept a user-supplied system ID and proceed without further checks that the user should have access to that system. As a result, any authenticated user can access these routes for any system if they know the system's ID. System IDs are random 15 character alphanumeric strings, and are not exposed to all users. However, it is theoretically possible for an authenticated user to enumerate a valid system ID via web API. To use the containers endpoints, the user would also need to enumerate a container ID, which is 12 digit hexadecimal string. This vulnerability is fixed in 0.18.7.

Improper Input Validation vulnerability in Apache Tomcat due to an incomplete fix of CVE-2025-66614. This issue affects Apache Tomcat: from 11.0.15 through 11.0.19, from 10.1.50 through 10.1.52, from 9.0.113 through 9.0.115. Users are recommended to upgrade to version 11.0.20, 10.1.53 or 9.0.116, which fix the issue.

Padding Oracle vulnerability in Apache Tomcat's EncryptInterceptor with default configuration. This issue affects Apache Tomcat: from 11.0.0-M1 through 11.0.18, from 10.0.0-M1 through 10.1.52, from 9.0.13 through 9..115, from 8.5.38 through 8.5.100, from 7.0.100 through 7.0.109. Users are recommended to upgrade to version 11.0.19, 10.1.53 and 9.0.116, which fixes the issue.

CLIENT_CERT authentication does not fail as expected for some scenarios when soft fail is disabled vulnerability in Apache Tomcat, Apache Tomcat Native. This issue affects Apache Tomcat: from 11.0.0-M1 through 11.0.18, from 10.1.0-M7 through 10.1.52, from 9.0.83 through 9.0.115; Apache Tomcat Native: from 1.1.23 through 1.1.34, from 1.2.0 through 1.2.39, from 1.3.0 through 1.3.6, from 2.0.0 through 2.0.13. Users are recommended to upgrade to version Tomcat Native 1.3.7 or 2.0.14 and Tomcat 11.0.20, 10.1.53 and 9.0.116, which fix the issue.

Configured cipher preference order not preserved vulnerability in Apache Tomcat. This issue affects Apache Tomcat: from 11.0.16 through 11.0.18, from 10.1.51 through 10.1.52, from 9.0.114 through 9.0.115. Users are recommended to upgrade to version 11.0.20, 10.1.53 or 9.0.116, which fix the issue.

A vulnerability was found in FoundationAgents MetaGPT up to 0.8.1. Impacted is the function get_mime_type of the file metagpt/utils/common.py. The manipulation results in os command injection. The attack can be executed remotely. The exploit has been made public and could be used. The project was informed of the problem early through a pull request but has not reacted yet.

Occasional URL redirection to untrusted Site ('Open Redirect') vulnerability in Apache Tomcat via the LoadBalancerDrainingValve. This issue affects Apache Tomcat: from 11.0.0-M1 through 11.0.18, from 10.1.0-M1 through 10.1.52, from 9.0.0.M23 through 9.0.115, from 8.5.30 through 8.5.100. Other, unsupported versions may also be affected Users are recommended to upgrade to version 11.0.20, 10.1.53 or 9.0.116, which fix the issue.