PraisonAI: Unauthenticated Allow-List Manipulation Bypasses Agent Tool Approval Safety Controls

SummaryThe gateway's /api/approval/allow-list endpoint permits unauthenticated modification of the tool approval allowlist when no authtoken is configured the default. By adding dangerous tool names e.g., shellexec, filewrite to the allowlist, an attacker can cause the ExecApprovalManager to...

PraisonAIAgents: Path Traversal via Unvalidated Glob Pattern in list_files Bypasses Workspace Boundary

SummaryThe listfiles tool in FileTools validates the directory parameter against workspace boundaries via validatepath, but passes the pattern parameter directly to Path.glob without any validation. Since Python's Path.glob supports .. path segments, an attacker can use relative path traversal in...

PraisonAIAgents: Arbitrary File Read via read_skill_file Missing Workspace Boundary and Approval Gate

Summaryreadskillfile in skilltools.py allows reading arbitrary files from the filesystem by accepting an unrestricted skillpath parameter. Unlike filetools.readfile which enforces workspace boundary confinement, and unlike runskillscript which requires critical-level approval, readskillfile has...

PraisonAI Vulnerable to Code Injection and Protection Mechanism Failure

PraisonAI's AST-based Python sandbox can be bypassed using type.getattribute trampoline, allowing arbitrary code execution when running untrusted agent code. DescriptionThe executecodedirect function in praisonaiagents/tools/pythontools.py uses AST filtering to block dangerous Python attributes...

PraisonAI Vulnerable to Decompression Bomb DoS via Recipe Bundle Extraction Without Size Limits

SummaryThe safeextractall 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...

PraisonAIAgents: Environment Variable Secret Exfiltration via os.path.expandvars() Bypassing shell=False in Shell Tool

SummaryThe executecommand function in shelltools.py calls os.path.expandvars on every command argument at line 64, manually re-implementing shell-level environment variable expansion despite using shell=False line 88 for security. This allows exfiltration of secrets stored in environment variable...

PraisonAI has Memory State Leakage and Path Traversal in MultiAgent Context Handling

SummaryThe MultiAgentLedger and MultiAgentMonitor components in the provided code exhibit vulnerabilities that can lead to context leakage and arbitrary file operations. Specifically:1. Memory State Leakage via Agent ID Collision: The MultiAgentLedger uses a dictionary to store ledgers by agent I...

LangChain has incomplete f-string validation in prompt templates

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 cou...

FoundationAgents MetaGPT vulnerable to eval injection

A flaw has been found in FoundationAgents MetaGPT up to 0.8.1. This vulnerability affects the function ActionNode.xmlfill of the file metagpt/actions/actionnode.py of the component XML Handler. Executing a manipulation can lead to improper neutralization of directives in dynamically evaluated cod...

MetaGPT has an Injection issue

A vulnerability was detected in FoundationAgents MetaGPT up to 0.8.1. This affects the function checksolution of the component HumanEvalBenchmark/MBPPBenchmark. Performing a manipulation results in code injection. The attack may be initiated remotely. The exploit is now public and may be used. Th...

FoundationAgents MetaGPT vulnerable to os command injection via the Terminal.run_command

A vulnerability has been found in FoundationAgents MetaGPT up to 0.8.1. This issue affects the function Terminal.runcommand in the library metagpt/tools/libs/terminal.py. The manipulation leads to os command injection. Remote exploitation of the attack is possible. The exploit has been disclosed ...

PraisonAI Vulnerable to Argument Injection into Cloud Run Environment Variables via Unsanitized Comma in gcloud --set-env-vars

Summarydeploy.py constructs a single comma-delimited string for the gcloud rundeploy --set-env-vars argument by directly interpolating openaimodel,openaikey, and openaibase without validating that these values do notcontain commas. gcloud uses a comma as the key-value pair separator...

PraisonAI: Unauthenticated WebSocket Endpoint Proxies to Paid OpenAI Realtime API Without Rate Limits

SummaryThe /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...

FoundationAgents MetaGPT vulnerable to OS Command Injection in metagpt/tools/libs/terminal.py

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...

PraisonAI Has Unauthenticated SSE Event Stream that Exposes All Agent Activity in A2U Server

The A2U Agent-to-User event stream server in PraisonAI exposes all agent activity without authentication. This is a separate component from the gateway server fixed in CVE-2026-34952.The createa2uroutes function registers the following endpoints with NO authentication checks:- GET /a2u/info —...

FoundationAgents MetaGPT vulnerable to OS Command Injection in metagpt/utils/common.py

A vulnerability was found in FoundationAgents MetaGPT up to 0.8.1. Impacted is the function getmimetype 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...

PraisonAI Vulnerable to Stored XSS via Unsanitized Agent Output in HTML Rendering (nh3 Not a Required Dependency)

SummaryThe Flask API endpoint in src/praisonai/api.py renders agent output as HTML without effective sanitization. The sanitizehtml 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...

PraisonAI has Template Injection in Agent Tool Definitions

SummaryDirect insertion of unescaped user input into template-rendering tools allows arbitrary code execution via specially crafted agent instructions. DetailsThe createagentcentrictools function returns tools like acpcreatefile that process file content using template rendering. When user input...

stata-mcp has insufficient validation of user-supplied Stata do-file content that can lead to command execution

A vulnerability was identified in stata-mcp prior to v1.13.0 where insufficient validation of user-supplied Stata do-file content can lead to command execution...

pyload-ng has a WebUI JSON permission mismatch that lets ADD/DELETE users invoke MODIFY-only actions

SummarySeveral WebUI JSON endpoints enforce weaker permissions than the core API methods they invoke. This allows authenticated low-privileged users to execute MODIFY operations that should be denied by pyLoad's own permission model.Confirmed mismatches:- ADD user can reorder packages/files...

lightrag-hku: JWT Algorithm Confusion Vulnerability

SummaryThe LightRAG API is vulnerable to a JWT algorithm confusion attack where an attacker can forge tokens by specifying 'alg': 'none' in the JWT header. Since the jwt.decode call does not explicitly deny the 'none' algorithm, a crafted token without a signature will be accepted as valid, leadi...

PraisonAI recipe registry publish path traversal allows out-of-root file write

SummaryPraisonAI's recipe registry publish endpoint writes uploaded recipe bundles to a filesystem path derived from the bundle's internal manifest.json before it verifies that the manifest name and version match the HTTP route. A malicious publisher can place ../ traversal sequences in the bundl...

Lupa has a Sandbox escape and RCE due to incomplete attribute_filter enforcement in getattr / setattr

SummaryThe attributefilter in the Lupa library is intended to restrict access to sensitive Python attributes when exposing objects to Lua.However, the filter is not consistently applied when attributes are accessed through built-in functions like getattr and setattr. This allows an attacker to...

OpenViking contains a missing authorization vulnerability in the task polling endpoints

OpenViking versions prior to 0.3.3 contain a missing authorization vulnerability in the task polling endpoints that allows unauthorized attackers to enumerate or retrieve background task metadata created by other users. Attackers can access the /api/v1/tasks and /api/v1/tasks/taskid routes withou...

rfc3161-client Has Improper Certificate Validation

SummaryAn Authorization Bypass vulnerability in rfc3161-client's signature verification allows any attacker to impersonate a trusted TimeStamping Authority TSA. By exploiting a logic flaw in how the library extracts the leaf certificate from an unordered PKCS7 bag of certificates, an attacker can...

parisneo/lollms has an insufficient session expiration vulnerability

An insufficient session expiration vulnerability exists in the latest version of parisneo/lollms. The application fails to invalidate active sessions after a password reset, allowing an attacker to continue using an old session token. This issue arises due to the absence of logic to reject reques...

OpenEXR has a signed 32-bit Overflow in PIZ Decoder Leads to OOB Read/Write

Summaryinternalexrundopiz advances the working wavelet pointer with signed 32-bit arithmetic:cwavbuf += nx ny wcount;Because nx, ny, and wcount are int, a crafted EXR file can make this product overflow and wrap. The next channel then decodes from an incorrect address. The wavelet decode path...

OpenEXR: DWA Lossy Decoder Heap Out-of-Bounds Write

SummaryThe DWA lossy decoder constructs temporary per-component block pointers using signed 32-bit arithmetic. For a large enough width, the calculation overflows and later decoder stores operate on a wrapped pointer outside the allocated rowBlock backing store.This bug is reachable from the publ...

OpenEXR Makes Use of Uninitialized Memory

SummaryWhile fuzzing openexrexrcheckfuzzer, Valgrind reports a conditional branch depending on uninitialized data inside genericunpack. This indicates a use of uninitialized memory CWE-457. The issue is reproducible with the current OSS-Fuzz harness and a single-file PoC. DetailsEnvironment:-...

kedro-datasets has a path traversal vulnerability in PartitionedDataset that allows arbitrary file write

ImpactPartitionedDataset in kedro-datasets was vulnerable to path traversal. Partition IDs were concatenated directly with the dataset base path without validation. An attacker or malicious input containing .. components in a partition ID could cause files to be written outside the configured...

OpenEXR has use after free in PyObject_StealAttrString

SummaryThere is a use-after-free in PyObjectStealAttrString of pyOpenEXRold.cpp.This bug was found with ZeroPath. DetailsThe legacy adapter defines PyObjectStealAttrString that calls PyObjectGetAttrString to obtain a new reference, immediately decrefs it, and returns the pointer. Callers then pas...

OpenEXR has buffer overflow in PyOpenEXR_old's channels() and channel()

SummaryA memory safety bug in the legacy OpenEXR Python adapter the deprecated OpenEXR.InputFile wrapper allow crashes and likely code execution when opening attacker-controlled EXR files or when passing crafted Python objects.Integer overflow and unchecked allocation in InputFile.channel and...

PraisonAI Has Arbitrary File Write (Zip Slip) in Templates Extraction

The PraisonAI templates installation feature is vulnerable to a "Zip Slip" Arbitrary File Write attack. When downloading and extracting template archives from external sources e.g., GitHub, the application uses Python's zipfile.extractall without verifying if the files within the archive resolve...

OpenEXR has heap-buffer-overflow via signed integer underflow in ImfContextInit.cpp

SummaryA heap-buffer-overflow OOB read occurs in the istreamnonparallelread function in ImfContextInit.cpp when parsing a malformed EXR file through a memory-mapped IStream. A signed integer subtraction produces a negative value that is implicitly converted to sizet, resulting in a massive length...

PyBlade: SSTI/RCE via Bypassed AST Validation in sandbox.py

A vulnerability has been found in AntaresMugisho PyBlade 0.1.8-alpha/0.1.9-alpha. The affected element is the function issafeast of the file sandbox.py of the component AST Validation. Such manipulation leads to improper neutralization of special elements used in a template engine. The attack may...

pyLoad: Unprotected storage_folder enables arbitrary file write to Flask session store and code execution (Incomplete fix for CVE-2026-33509)

SummaryThe fix for CVE-2026-33509 GHSA-r7mc-x6x7-cqxx added an ADMINONLYOPTIONS set to block non-admin users from modifying security-critical config options. The storagefolder option is not in this set and passes the existing path restriction because the Flask session directory is outside both...

PraisonAI recipe registry pull path traversal writes files outside the chosen output directory

SummaryPraisonAI's recipe registry pull flow extracts attacker-controlled .praison tar archives with tar.extractall and does not validate archive member paths before extraction. A malicious publisher can upload a recipe bundle that contains ../ traversal entries and any user who later pulls that...

curl_cffi: Redirect-based SSRF leads to internal network access in curl_cffi (with TLS impersonation bypass)

Summarycurlcffi does not restrict requests to internal IP ranges, and follows redirects automatically via the underlying libcurl.Because of this, an attacker-controlled URL can redirect requests to internal services such as cloud metadata endpoints. In addition, curlcffi’s TLS impersonation featu...

LiteLLM: Privilege escalation via unrestricted proxy configuration endpoint

ImpactThe /config/update endpoint does not enforce admin role authorization. A user who is already authenticated into the platform can then use this endpoint to do the following: - Modify proxy configuration and environment variables - Register custom pass-through endpoint handlers pointing to...

Ajenti has an authorization bypass during custom package installation

ImpactAn authenticated user using the authusers plugin authentication method could install a custom package even if this user is not superuser. PatchesThis is fixed in the version 2.2.15. Users should upgrade to this version as soon as possible...

D-Tale: Remote Code Execution through redis/shelf storage

ImpactUsers hosting D-Tale publicly while using a redis or shelf storage layer could be vulnerable to remote code execution allowing attackers to run malicious code on the server. PatchesUsers should upgrade to version 3.22.0. WorkaroundsThere are no workarounds for versions 3.22.0...

OpenEXR: Heap information disclosure in PXR24 decompression via unchecked decompressed size (undo_pxr24_impl)

SummaryThe PXR24 decompression function undopxr24impl in OpenEXR internalpxr24.c ignores the actual decompressed size outSize returned by exruncompressbuffer and instead reads from the scratch buffer based solely on the expected size uncompressedsize derived from the header metadata.Additionally,...

LightRAG: Hardcoded JWT Signing Secret Allows Authentication Bypass

Subject: Security Vulnerability Report Hardcoded JWT Secret CVE-2026-30762Hi HKUDS team,I'm writing to report a security vulnerability I discovered in LightRAG v1.4.10. This has been assigned CVE-2026-30762 by MITRE.Vulnerability: Hardcoded JWT signing secretType: Improper Authentication...

web3.py: SSRF via CCIP Read (EIP-3668) OffchainLookup URL handling

Summaryweb3.py implements CCIP Read / OffchainLookup EIP-3668 by performing HTTP requests to URLs supplied by smart contracts in offchainlookuppayload"urls". The implementation uses these contract-supplied URLs directly after sender / data template substitution without any destination validation:...

OpenEXR: integer overflow to OOB write in uncompress_b44_impl()

SummaryThe B44/B44A decoder in OpenEXR reconstructs row pointers into a scratch buffer using int. When the channel width nx is large enough, the product y nx overflows int, causing the row pointer to wrap before the start of the scratch buffer. Subsequent memcpy calls then write decoded pixel...

pyLoad: SSRF in parse_urls API endpoint via unvalidated URL parameter

Vulnerability DetailsCWE-918: Server-Side Request Forgery SSRFThe parseurls API function in src/pyload/core/api/init.py line 556 fetches arbitrary URLs server-side via geturlurl pycurl without any URL validation, protocol restriction, or IP blacklist. An authenticated user with ADD permission can...

vLLM: Server-Side Request Forgery (SSRF) in `download_bytes_from_url `

SummaryA Server Side Request Forgery SSRF vulnerability in downloadbytesfromurl allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions.This can be used to target interna...

LTI JupyterHub Authenticator: Unbounded Memory Growth via Nonce Storage (Denial of Service)

SummaryThe LTI 1.1 validator stores OAuth nonces in a class-level dictionary that grows without bounds. Nonces are added before signature validation, so an attacker with knowledge of a valid consumer key can send repeated requests with unique nonces to gradually exhaust server memory, causing a...

PraisonAI Has SSRF in FileTools.download_file() via Unvalidated URL

SummaryFileTools.downloadfile in praisonaiagents validates the destination path but performs no validation on the url parameter, passing it directly to httpx.stream with followredirects=True. An attacker who controls the URL can reach any host accessible from the server including cloud metadata...

FastMCP has a Command Injection vulnerability - Gemini CLI

Server names containing shell metacharacters e.g., & can cause command injection on Windows when passed to fastmcp install claude-code or fastmcp install gemini-cli. These install paths use subprocess.run with a list argument, but on Windows the target CLIs often resolve to .cmd wrappers that are...