🚨 CVE-2026-15311
A vulnerability was identified in NousResearch hermes-agent up to 2026.5.29.2. Affected by this issue is the function MatrixAdapter._markdown_to_html of the file gateway/platforms/matrix.py of the component Matrix Adapter. Such manipulation leads to cross site scripting. The attack can be executed remotely. The exploit is publicly available and might be used. The pull request to fix this issue awaits acceptance.

🎖@cveNotify

🚨 CVE-2026-15317
A security flaw has been discovered in Sipeed PicoClaw up to 0.2.9. Affected by this vulnerability is the function WebFetchTool.Execute of the file pkg/tools/integration/web.go of the component Guarded Web Fetch Flow. The manipulation results in server-side request forgery. The attack can be executed remotely. The exploit has been released to the public and may be used for attacks. The reported GitHub issue was closed automatically due to inactivity.

🎖@cveNotify

🚨 CVE-2026-50180
Langroid is a framework for building large-language-model-powered applications. Prior to version 0.64.0, `SQLChatAgent` in `langroid` ships a `_validate_query` defense-in-depth layer whose `_DANGEROUS_SQL_PATTERNS` regex blocklist enumerates dangerous SQL primitives by specific function name. The list misses the canonical PostgreSQL filesystem-disclosure family `pg_read_file()`, `pg_stat_file()`, `pg_ls_logdir()`, `pg_ls_waldir()`, `pg_current_logfile()` (and similar `SELECT`-shaped functions in the same family). It also leaves SQL Server `OPENDATASOURCE` and SQLite `ATTACH '<file>' AS x` (DATABASE keyword omitted) unblocked. An attacker able to shape the LLM's generated SQL (directly via prompt input or transitively via prompt-injection in data the LLM ingests) can read arbitrary files from the PostgreSQL host through ordinary `SELECT` queries, even with the agent's strict default configuration (`allow_dangerous_operations=False`, `allowed_statement_types=['SELECT']`). The payloads survive the statement-type allowlist (each is a `SELECT`) and pass through the regex blocklist (none of the function names match), then reach the live SQLAlchemy engine via `SQLChatAgent.run_query`. Version 0.64.0 contains a patch for the issue.

🎖@cveNotify

🚨 CVE-2026-50181
Langroid is a framework for building large-language-model-powered applications. Prior to version 0.64.0, Langroid's `ReadFileTool` and `WriteFileTool` appear to treat `curr_dir` as the intended working-directory boundary for file operations. However, the tools only change the process working directory to `curr_dir` and then operate on the user-supplied `file_path` without resolving and enforcing that the final path remains inside `curr_dir`. As a result, a tool caller can supply path traversal sequences such as `../secret.txt` to read files outside the configured current directory, or `../written_by_tool.txt` to write files outside that directory. This can impact applications that expose Langroid file tools to an LLM agent, user-controlled tool call, or delegated coding/documentation agent while relying on `curr_dir` to restrict file access to a project/workspace directory. Version 0.64.0 patches the issue.

🎖@cveNotify

🚨 CVE-2026-54760
Langroid is a framework for building large-language-model-powered applications. Prior to version 0.65.1, the `SQLChatAgent` SQL-injection mitigation, with default `allow_dangerous_operations=False`, combines a raw-text regex blocklist (`_DANGEROUS_SQL_PATTERNS`) with a `sqlglot` SELECT-only statement allowlist. The blocklist entries that target callable functions require the function name to be immediately followed by `\s*\(`. PostgreSQL accepts the same call with the name separated from `(` by a quoted identifier, an inline comment, or schema qualification. These forms evade the regex, still parse as `SELECT`, and execute the same PostgreSQL function. This restores the `pg_read_file` server-side file-read primitive that the prior CVE-2026-25879 / GHSA-pmch-g965-grmr fix was meant to block: the parent advisory fixed a missing `pg_read_file` blocklist entry, while this report shows that the added regex is bypassable. Version 0.65.1 fixes the issue.

🎖@cveNotify

🚨 CVE-2026-54769
Langroid is a framework for building large-language-model-powered applications. Versions prior to 0.65.2 are vulnerable to a critical Sandbox Escape leading to Remote Code Execution (RCE) in its `TableChatAgent` and `VectorStore` capabilities. When these agents evaluate LLM-generated tool messages with `full_eval=True`, they attempt to sandbox the execution by explicitly setting `locals` to an empty dictionary `{}` inside Python's `eval()` function. However, this relies on an incomplete understanding of Python's execution model. Because `__builtins__` is not explicitly scrubbed from the `globals` dictionary mapping, Python implicitly injects all built-ins during execution, granting full access to functions like `__import__('os').system()`. Since `TableChatAgent.pandas_eval()` executes external LLM outputs natively, this bypass permits any attacker providing prompt payload to achieve unauthenticated RCE on the host system. Version 0.65.2 patches the issue.

🎖@cveNotify

🚨 CVE-2026-54771
Langroid is a framework for building large-language-model-powered applications. Prior to version 0.65.3, a Langroid application exposing a chat interface to untrusted users may allow direct tool invocation via raw JSON payloads, even when tools are registered with `use=False, handle=True`. Version 0.65.3 fixes the issue.

🎖@cveNotify

🚨 CVE-2026-55615
Langroid is a framework for building large-language-model-powered applications. Prior to version 0.65.5, Neo4jChatAgent passes LLM-generated Cypher queries straight to the Neo4j driver with no validation, no statement-type allowlist, and no opt-out gate. The query text is influenceable by prompt injection (direct user input or indirect content the agent reads back via RAG), so an attacker who can influence the prompt can read or destroy all graph data and, when APOC or dbms.security procedures are enabled on the server, achieve OS-command and filesystem access. This is the same defect class and threat model as the SQLChatAgent prompt-to-SQL-to-RCE issue fixed in version 0.63.0 (CVE-2026-25879); that fix did not extend to the neo4j module. Version 0.65.5 contains a fix for the neo4j module.

🎖@cveNotify

🚨 CVE-2026-44241
Micronaut Framework is a JVM-based full stack Java framework designed for building modular, easily testable JVM applications. From 4.3.0 to before 4.10.22, 3.10.6, and 3.8.14, TimeConverterRegistrar caches DateTimeFormatter instances in an unbounded ConcurrentHashMap<String, DateTimeFormatter> whose key is derived from the @Format annotation pattern concatenated with the locale from the HTTP Accept-Language header. Because Locale.forLanguageTag() accepts arbitrary BCP 47 private-use extensions (en-x-a001, en-x-a002, …), an unauthenticated attacker can generate an unlimited number of unique cache keys by sending requests with novel locale tags, growing the cache until heap memory is exhausted and the JVM crashes. This vulnerability is fixed in 4.10.22, 3.10.6, and 3.8.14.

🎖@cveNotify

🚨 CVE-2026-13757
A flaw was found in p11-kit. The RPC message attribute parsing functions p11_rpc_message_get_attribute() and p11_rpc_message_get_attribute_array_value() form a mutually-recursive call chain with no recursion depth limit when processing nested CKA_WRAP_TEMPLATE, CKA_UNWRAP_TEMPLATE, and CKA_DERIVE_TEMPLATE attributes. An unauthenticated attacker with local access to the p11-kit RPC Unix domain socket can send a specially crafted request with deeply nested template attributes, causing stack exhaustion and crashing the p11-kit server process and its dependent services.

🎖@cveNotify

🚨 CVE-2026-15318
A weakness has been identified in Sipeed PicoClaw up to 0.2.9. Affected by this issue is some unknown functionality of the file pkg/channels/mqtt/mqtt.go of the component MQTT Channel Handler. This manipulation of the argument client_id causes incorrect authorization. The attack is possible to be carried out remotely. The exploit has been made available to the public and could be used for attacks. The reported GitHub issue was closed automatically due to inactivity.

🎖@cveNotify

🚨 CVE-2026-50082
The Aqara Cloud Developer Portal (developer.aqara.com) issued a developer token to any email address supplied by the attacker. This is an instance of "CWE-306: Missing Authentication for Critical Function" with an estimated CVSS of CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N (6.5 Medium). When combined with CVE-2026-50083, CVE-2026-50084, and CVE-2026-50085, any otherwise-unauthenticated attacker could execute a full takeover of affected devices.

🎖@cveNotify

🚨 CVE-2026-54401
A malicious actor with access to the network and low privileges could exploit a Server-Side Request Forgery (SSRF) to escalate privileges within such UniFi OS devices or instances.

🎖@cveNotify

🚨 CVE-2026-54402
A malicious actor with access to the network and low privileges could exploit an Improper Input Validation vulnerability found in UniFi OS to execute a Command Injection on the host device.

🎖@cveNotify

🚨 CVE-2026-54403
A malicious actor with access to the network could exploit a Path Traversal vulnerability found in certain devices running UniFi OS to bypass authentication of such UniFi OS devices or instances.

🎖@cveNotify

🚨 CVE-2026-54404
A malicious actor with access to the network and low privileges could exploit a series of authenticated SQL Injection vulnerabilities found in UniFi OS to escalate privileges within such UniFi OS devices or instances.

🎖@cveNotify

🚨 CVE-2026-55112
A malicious actor with access to the network and low privileges and under certain conditions could exploit an Improper Access Control vulnerability found in UniFi OS with UniFi Protect Application to escalate privileges on the host device.

🎖@cveNotify

🚨 CVE-2026-12590
Impact: In body-parser versions prior to 1.20.6 (1.x line) and 2.3.0 (2.x line), when the parser is configured with an invalid limit option value such as an unparseable string or NaN, bytes.parse returns null and the request body size check is silently skipped. Applications that rely on limit as their primary safeguard against oversized request bodies will accept arbitrarily large payloads, leading to excessive memory and CPU usage and denial of service. Patches: This issue is fixed in body-parser 1.20.6 and 2.3.0. After the fix, invalid limit values throw a clear error at parser construction time instead of silently disabling enforcement, while null and undefined continue to fall back to the default limit of 100kb. Workarounds: Validate the limit value before passing it to body-parser. For example, parse the value at startup and reject any configuration where the result is null or a non-finite number.

🎖@cveNotify

🚨 CVE-2026-59225
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.8.12 before 0.10.0, an authenticated non-admin user with read access to an arena wrapper model can reach a restricted underlying model through task endpoints such as /api/v1/tasks/moa/completions. The normal chat route resolves arena models before the final chat dispatch and therefore re-checks the selected underlying model. The task routes call utils.chat.generate_chat_completion() directly. In that direct path, arena fallback resolution happens after the wrapper access check and then recurses with bypass_filter=True, skipping the selected submodel's access check. This issue is fixed in version 0.10.0.

🎖@cveNotify

🚨 CVE-2026-59226
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.0 before 0.10.0, execute_automation rehydrated automation owners without rechecking that they were still active or still had features.automations, and check_model_access only enforced private-model grants for the exact user role, allowing deactivated pending users to continue scheduled model execution. This issue is fixed in version 0.10.0.

🎖@cveNotify

🚨 CVE-2026-59227
Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.8.11 before 0.10.0, POST /api/v1/images/edit required only a verified account and did not enforce the global image-edit switch or the per-user image-generation permission, allowing a non-admin user to invoke server-side image editing with administrator-configured provider credentials. This issue is fixed in version 0.10.0.

🎖@cveNotify