CVE
CVE-2026-116692, CVE-2026-116690, CVE-2026-116691, CVE-2026-116684, CVE-2026-116685, CVE-2026-116686, CVE-2026-115419, CVE-2026-115413, CVE-2026-115522, CVE-2026-115883, CVE-2026-118668, CVE-2026-118667, CVE-2026-115885, CVE-2026-116575, CVE-2026-115886, CVE-2026-118623, CVE-2026-118627, CVE-2026-118628
CWE
CWE-94, CWE-22, CWE-59, CWE-532, CWE-488, CWE-668, CWE-706, CWE-862, CWE-346, CWE-918, CWE-502
Affected Surface
- PyPI `mlflow` versions `>= 3.12.0rc0, < 3.17.0` (`CVE-2026-116692` CVSS 9.8 RCE and `CVE-2026-115886` SSRF), `>= 3.10.0, < 3.17.0` (`CVE-2026-116690` arbitrary file read), and `>= 2.19.0, < 3.17.0` (`CVE-2026-116691` CrewAI autolog API-key disclosure), plus ZDI advisories `CVE-2026-118623` (`ZDI-26-839`), `CVE-2026-118627` (`ZDI-26-843`), and `CVE-2026-118628` (`ZDI-26-844`)
- PyPI `mcp` (Model Context Protocol Python SDK) versions `>= 1.8.0, < 1.30.0` and `>= 2.0.0, < 2.2.0` (`CVE-2026-116684` / `GHSA-w4fh-qvv9-3v23`, `CVE-2026-116685` / `GHSA-5h93-6whr-6q8j`, `CVE-2026-116686` / `GHSA-rwrf-2pqf-9j8j`)
- PyPI `strands-agents < 1.48.0` (`GHSA-3w55-933p-8pv5` symlink LPE and `GHSA-763v-5pjm-fx36` world-readable `/tmp/strands/sessions` storage) and `strands-agents < 1.43.0` / npm `@strands-agents/sdk < 1.5.0` running multi-caller Agent2Agent (`A2AServer`) endpoints (`CVE-2026-115419` / `GHSA-4v6f-7vjh-6x4m`)
- PyPI `awslabs.aws-api-mcp-server >= 1.3.0, < 1.6.0` using `streamable-http` OAuth authentication (`CVE-2026-115413` / `GHSA-4prq-p2x5-w9fh`), and PyPI `google-adk >= 1.19.0, < 1.39.1` and `>= 2.0.0a1, < 2.7.0` sharing a `FileArtifactService` root across multiple apps (`CVE-2026-115522`)
- PyPI `langgraph-api >= 0.1.2, < 0.15.0` (`CVE-2026-118668` / `GHSA-4hm6-w6qq-w73v`), `langgraph-sdk >= 0.4.0, < 0.4.6` / `langgraph >= 1.2.3, < 1.2.14` (`CVE-2026-118667`), `langgraph-cli >= 0.4.15, < 0.4.32` (`CVE-2026-115885`), `langgraph-checkpoint-postgres < 3.1.3` / `langgraph-checkpoint-sqlite < 3.1.2` (`CVE-2026-115883`), and PyPI `mem0ai < 2.0.18` / npm `mem0ai < 3.1.6` (`CVE-2026-116575` / `GHSA-5324-3w58-qpp2`)
- PyPI `fastmcp` and `fastmcp-slim` `>= 3.0.0, < 3.4.10` and `>= 4.0.0, < 4.1.2` using component `auth=` checks with transforms (`GHSA-7xgj-8392-5q4r`) or `ClientCredentialsOAuthProvider` / `PrivateKeyJWTOAuthProvider` (`GHSA-vqqp-x5qm-5qp4`)
Update, 5 November 2026: On 4 November (00:17 UTC), Trend Micro’s Zero Day Initiative (ZDI) and NVD published three additional advisories for MLflow: CVE-2026-118623 (ZDI-26-839, CVSS 7.8 High, untrusted pickle deserialization in PickleEvaluationArtifact._save / _load when evaluating untrusted artifacts), CVE-2026-118627 (ZDI-26-843, CVSS 7.8 High, untrusted pickle deserialization in mlflow.shap._load_pyfunc / load_explainer when loading SHAP explainer models), and CVE-2026-118628 (ZDI-26-844, CVSS 8.2 High, unauthenticated server-side request forgery in the /get-artifacts handler via unvalidated artifact URIs).
Update, 29 October 2026: Late on 28 October (21:05 to 21:17 UTC), four additional advisories were published for LangGraph and Mem0 (mem0ai):
- LangGraph API JavaScript custom-auth Store namespace bypass (
CVE-2026-118668/GHSA-4hm6-w6qq-w73v, CVSS 8.6 High, fixed inlanggraph-api 0.15.0): Inlanggraph-api >= 0.1.2, < 0.15.0, when a deployment used JavaScript or TypeScript custom authentication and its Store authorization handler restricted tenants by rewriting the requested namespace rather than rejecting the request outright, the server failed to carry the rewritten namespace across the JS/Python runtime boundary before executing the Store operation. Any authenticated user could read, search, overwrite, or delete another user’s Store data. - LangGraph SDK v3 streaming path traversal (
CVE-2026-118667/GHSA-3fx2-cqw9-xr3c, Medium, fixed inlanggraph-sdk 0.4.6andlanggraph 1.2.14) andlanggraph deploy --verbosepush-token leak (CVE-2026-115885/GHSA-8prr-f582-jqq3, Medium, fixed inlanggraph-cli 0.4.32): Inlanggraph-sdk0.4.0through0.4.5, the v3 streaming client (get_treeand_fetch_state) interpolatedthread_idandassistant_idinto URL paths without calling_quote_path_param(commitd89c7e100e), letting relative path segments redirect authenticated GET requests to/store/items?.... Meanwhile, inlanggraph-cli0.4.15through0.4.31, runninglanggraph deploy --verboseechoed thedocker login --password-stdinstdin stream, printing the temporary registry push token into CI logs (commit1afaca35a0). - Mem0
_build_session_scopedelimiter collision and cross-tenant conversation leak (CVE-2026-116575/GHSA-5324-3w58-qpp2, CVSS 6.3 Medium, fixed in PyPImem0ai 2.0.18and npmmem0ai 3.1.6): Bothmem0/memory/main.py(_build_session_scope) andmem0-ts/src/oss/src/memory/index.ts(buildSessionScope) joinedagent_id,run_id, anduser_idinto"&".join(f"{key}={val}" ...)without forbidding or escaping&and=. A caller who controlled onlyrun_idcould passrun_id="proj-x&user_id=alice"to collide byte-for-byte with{"user_id": "alice", "run_id": "proj-x"}, causingMemory.add()to read Alice’s last 10 raw messages into the attacker’s LLM fact-extraction prompt and overwrite or evict Alice’s conversation history (commit71f2ebefa3).
Update, 28 October 2026: Between 27 and 28 October, ten additional advisories were published for MLflow 3.17.0, the Model Context Protocol (MCP) Python SDK (mcp 1.30.0 and 2.2.0), Amazon Strands Agents SDK 1.48.0, AWS API MCP Server 1.6.0, and Google Agent Development Kit (google-adk 1.39.1 and 2.7.0):
- Strands Agents
FileSessionManager/tmpsymlink LPE and world-readable session storage (GHSA-3w55-933p-8pv5, CVSS 7.1 High, andGHSA-763v-5pjm-fx36, Medium, fixed instrands-agents 1.48.0): Instrands-agents < 1.48.0,FileSessionManagerdefaultedstorage_dirtoos.path.join(tempfile.gettempdir(), "strands/sessions")(/tmp/strands/sessionson Linux and macOS), created directories and JSON files under the default022umask (0755/0644), and wrote updates viaopen(f"{path}.tmp", "w")followed byos.replace()withoutO_NOFOLLOW. Any local user on a shared developer workstation, notebook server, or CI runner could read persisted prompts, tool outputs, and secrets (GHSA-763v-5pjm-fx36), pre-create a symlink atsession.json.tmpto overwrite arbitrary files owned by the victim user, or tamper with session files for persistent prompt injection (GHSA-3w55-933p-8pv5, commit8658b9e0d5). Version1.48.0moves the default directory to~/.strands/sessions/with mode0700. - AWS API MCP Server OAuth token audience bypass (
CVE-2026-115413/GHSA-4prq-p2x5-w9fh, CVSS 7.8 High, fixed inawslabs.aws-api-mcp-server 1.6.0): Whenawslabs.aws-api-mcp-server(>= 1.3.0, < 1.6.0) ran withAWS_API_MCP_TRANSPORT="streamable-http"andAUTH_TYPE="oauth", it verified the bearer token’s JWKS signature,iss, andexp, but never checkedaudorclient_id/azp. Any token issued by the same identity provider (such as another app client in the same Amazon Cognito user pool) was accepted and allowed the caller to execute AWS CLI commands under the server’s host IAM credentials. Version1.6.0(commitb97eefc7c2) requires setting eitherAUTH_AUDIENCEorAUTH_CLIENT_IDalongsideAUTH_ISSUERandAUTH_JWKS_URI. - Google Agent Development Kit (
google-adk) cross-appFileArtifactServiceleak (CVE-2026-115522, CVSS v4.0 5.3 Medium, fixed in1.39.1and2.7.0): Ingoogle-adk1.19.0before1.39.1and2.0.0a1before2.7.0,FileArtifactServicestored artifacts underroot/users/{user_id}/...and droppedapp_namefrom the path entirely, allowing one application to list, read, overwrite, or delete artifacts belonging to another application sharing the same artifact root (commitf72f0db58c). - MLflow unauthenticated third-party scorer artifact RCE (
CVE-2026-116692/GHSA-26p8-2jq9-3vq9, CVSS 9.8 Critical): Inmlflow3.12.0rc0through3.16.x,Scorer.model_validate()checkedmodule_path == m or module_path.startswith(m + ".")againstTHIRD_PARTY_SCORER_ALLOWED_MODULES(mlflow.genai.scorers.trulens,ragas,deepeval,phoenix) and then ranimportlib.import_module(module_path). By creating a file-backed experiment withartifact_location: "file:///usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens", uploading a.pyfile as a run artifact, and callingPOST /ajax-api/3.0/mlflow/scorer/invokewithmodule: "mlflow.genai.scorers.trulens.<run_id>.artifacts.<name>", an unauthenticated caller executed arbitrary Python code asrooton the officialghcr.io/mlflow/mlflowcontainer image (commita732443e69). - MLflow prompt model-version arbitrary file read (
CVE-2026-116690/GHSA-m65p-9j7q-6w33, CVSS 7.5 High) and CrewAI autolog API-key leakage (CVE-2026-116691/GHSA-j7qq-vr8c-xvw3, CVSS 6.5 Medium): Inmlflow3.10.0through3.16.1, theCreateModelVersionprompt branch (mlflow.prompt.is_prompttag) rejectedfile://URIs and leading/paths but skipped_validate_non_local_source_contains_relative_paths()wheneverparsed.schemewas empty, allowingsource: "../../../etc"or"."to serve arbitrary server files (includingbasic_auth.db) via/api/2.0/mlflow/model-versions/get-artifact(commit81730a2ce4). Separately, inmlflow2.19.0through3.16.x,mlflow.crewai.autolog()serializedAgent.__dict__into span attributes andmlflow.spanInputs, exporting plaintextapi_key='sk-proj-...'strings from CrewAILLMobjects across 5 span locations (commit20927930d2). - MCP Python SDK (
mcp1.30.0/2.2.0):CVE-2026-116684(GHSA-w4fh-qvv9-3v23, CVSS 6.8 Medium):BearerAuthBackendnever checkedAccessToken.resource(RFC 8707 audience) againstAuthSettings.resource_server_url. Upgrading alone does not enable the fix: you must setvalidate_token_resource=TrueinAuthSettingsand return the token’s audience inAccessToken.resourcefromverify_token()(commit0c9136841f).CVE-2026-116686(GHSA-rwrf-2pqf-9j8j, CVSS 6.5 Medium): When validatingcall_toolresults against a server’soutputSchema,ClientSession._validate_tool_result()calledjsonschema.validate()without an isolatedreferencing.Registry(), allowing a malicious MCP server to placehttp://,https://, orfile://$refURIs inoutputSchemato trigger synchronous SSRF, local file probing, or event-loop hangs on the client (commit5ebdfed406).CVE-2026-116685(GHSA-5h93-6whr-6q8j, CVSS 5.9 Medium):create_mcp_http_client()defaulted tofollow_redirects=True, causing Streamable HTTP, SSE, and OAuth client flows to forward custom credential headers (X-API-Key) and307/308request bodies across cross-origin redirects (commit2106335745).
TL;DR
Between 22 and 23 October 2026, four major AI agent and evaluation frameworks (Amazon Strands Agents SDK, LangGraph, Prefect FastMCP, and MLflow) disclosed five vulnerabilities affecting multi-tenant session isolation, MCP tool authorization, OAuth client-credential binding, and tracking-server egress:
- Amazon Strands Agents SDK (
CVE-2026-115419/GHSA-4v6f-7vjh-6x4m, CVSS 8.1 High):A2AServerandStrandsA2AExecutoraccepted a singleAgentinstance and reused its in-memory conversation state and streaming state across every caller’s Agent2Agent (A2A)context_id. Any caller could read another tenant’s conversation history or inject persistent prompt instructions into subsequent sessions. - LangGraph Checkpoint Postgres & SQLite (
CVE-2026-115883/GHSA-6hmw-9432-67qq, CVSS 6.0 Medium):PostgresStoreandSqliteStoreserialize a namespace tuple("tenant", "bob")by joining labels with a period (".".join(namespace)->"tenant.bob").BaseStoreonly rejected.inside labels onput()/aput(), leavingget(),search(),delete(),list_namespaces(), andbatch()/abatch()(includingPutOpitems insidebatch()) unchecked. Passing("tenant.bob",)read, overwrote, or deleted("tenant", "bob"). - Prefect FastMCP (
GHSA-7xgj-8392-5q4r, CVSS 6.8 Medium, andGHSA-vqqp-x5qm-5qp4, CVSS 5.9 Medium): First, when aToolTransformor customTransformadded tags or modified metadata that anAuthCheck(such asrestrict_tag("admin", scopes=["admin"])) relied on,list_tools()hid the tool, but directcall_tool()/get_resource()lookups checked the pre-transform component or returned the transformed component without running_component_auth_allows(). Second,ClientCredentialsOAuthProviderandPrivateKeyJWTOAuthProvidernever exposed or forwarded theissuer=parameter to the underlying MCP SDK, so a compromised or user-supplied MCP server could advertise its own OAuth token endpoint and capture the client’sclient_idandclient_secret. - MLflow (
CVE-2026-115886/GHSA-2wqp-q3c6-ccj5, CVSS 6.5 Medium): The unauthenticatedPOST /ajax-api/3.0/mlflow/scorer/invokeendpoint deserializedthird_party_scorer_dataviaScorer.model_validate_json(), reflectively instantiatingmlflow.genai.scorers.trulensclasses and forwarding rawkwargs(api_base,base_url,api_key,completion_kwargs) straight intoLiteLLM(), allowing unauthenticated SSRF with caller-controlled headers.
What to do now:
- Upgrade
strands-agentsto>= 1.43.0(@strands-agents/sdkto>= 1.5.0) and migrateA2AServer(agent=...)calls toA2AServer(agent_factory=lambda context_id: ...). - Upgrade
langgraph-checkpoint-postgresto>= 3.1.3,langgraph-checkpoint-sqliteto>= 3.1.2, andlanggraph-checkpointto>= 4.3.0. - Upgrade
fastmcp/fastmcp-slimto>= 4.1.2(or>= 3.4.10on 3.x), and explicitly passissuer="https://..."toClientCredentialsOAuthProviderandPrivateKeyJWTOAuthProvider. - Upgrade
mlflowto>= 3.17.0.
Affected packages and projects
| Ecosystem | Package | CVE / GHSA | Severity | Affected versions | Fixed versions | Action required beyond upgrade? |
|---|---|---|---|---|---|---|
| PyPI / npm | strands-agents, @strands-agents/sdk | CVE-2026-115419 (GHSA-4v6f-7vjh-6x4m) | CVSS 8.1 High | strands-agents < 1.43.0; @strands-agents/sdk < 1.5.0 | 1.43.0 (PyPI), 1.5.0 (npm) | Yes (recommended): pass agent_factory= (agentFactory) to A2AServer so each context_id gets its own Agent and session_manager |
| PyPI | strands-agents | GHSA-3w55-933p-8pv5, GHSA-763v-5pjm-fx36 | CVSS 7.1 High / Medium | < 1.48.0 | 1.48.0 (1.62.0 current) | Remove legacy /tmp/strands/sessions directories on shared hosts |
| PyPI | awslabs.aws-api-mcp-server | CVE-2026-115413 (GHSA-4prq-p2x5-w9fh) | CVSS 7.8 High | >= 1.3.0, < 1.6.0 | 1.6.0 | Yes: configure AUTH_AUDIENCE or AUTH_CLIENT_ID when AUTH_TYPE=oauth |
| PyPI | google-adk | CVE-2026-115522 | CVSS v4.0 5.3 Medium | >= 1.19.0, < 1.39.1; >= 2.0.0a1, < 2.7.0 | 1.39.1, 2.7.0 (2.13.0 current) | Pre-upgrade artifacts under root/users/ remain readable until migrated or deleted |
| PyPI | langgraph-checkpoint-postgres, langgraph-checkpoint-sqlite, langgraph-checkpoint | CVE-2026-115883 (GHSA-6hmw-9432-67qq) | CVSS 6.0 Medium | langgraph-checkpoint-postgres < 3.1.3; langgraph-checkpoint-sqlite < 3.1.2; langgraph-checkpoint < 4.3.0 | 3.1.3, 3.1.2, 4.3.0 | Custom external stores that join namespaces with . should call validate_op_namespace(op) |
| PyPI | fastmcp, fastmcp-slim | GHSA-7xgj-8392-5q4r | CVSS 6.8 Medium | >= 3.0.0, < 3.4.10; >= 4.0.0, < 4.1.2 | 3.4.10, 4.1.2 | No |
| PyPI | fastmcp, fastmcp-slim | GHSA-vqqp-x5qm-5qp4 | CVSS 5.9 Medium | >= 4.0.0a1, < 4.1.2 | 4.1.2 (requires mcp >= 2.2.0) | Yes: pass issuer="https://..." to ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider, and rotate secrets exposed to untrusted MCP URLs |
| PyPI | mcp | CVE-2026-116684 (GHSA-w4fh-qvv9-3v23) | CVSS 6.8 Medium | >= 1.10.0, < 1.30.0; >= 2.0.0, < 2.2.0 | 1.30.0, 2.2.0 | Yes: set validate_token_resource=True in AuthSettings and populate AccessToken.resource in verify_token() |
| PyPI | mcp | CVE-2026-116686 (GHSA-rwrf-2pqf-9j8j), CVE-2026-116685 (GHSA-5h93-6whr-6q8j) | CVSS 6.5 / 5.9 Medium | >= 1.8.0, < 1.30.0; >= 2.0.0, < 2.2.0 | 1.30.0, 2.2.0 | No (cross-origin redirects now fail unless endpoint URL is updated to final target) |
| PyPI | mlflow | CVE-2026-116692 (GHSA-26p8-2jq9-3vq9) | CVSS 9.8 Critical | >= 3.12.0rc0, < 3.17.0 | 3.17.0 (3.17.2 current) | Audit writable site-packages/mlflow/genai/scorers/* directories for rogue run artifact subtrees |
| PyPI | mlflow | CVE-2026-116690 (GHSA-m65p-9j7q-6w33) | CVSS 7.5 High | >= 3.10.0, < 3.17.0 | 3.17.0 | Audit registered model versions with mlflow.prompt.is_prompt tags for relative source paths |
| PyPI | mlflow | CVE-2026-116691 (GHSA-j7qq-vr8c-xvw3) | CVSS 6.5 Medium | >= 2.19.0, < 3.17.0 | 3.17.0 | Yes: rotate LLM provider keys used with mlflow.crewai.autolog() and purge recorded spans |
| PyPI | mlflow | CVE-2026-115886 (GHSA-2wqp-q3c6-ccj5) | CVSS 6.5 Medium | >= 3.12.0rc0, < 3.17.0 | 3.17.0 | No (previously stored scorer rows with destination kwargs are also blocked on read) |
Technical analysis
1. Strands Agents SDK (CVE-2026-115419): shared Agent state across A2A context_ids
In Amazon’s Strands Agents SDK (strands-agents on PyPI and @strands-agents/sdk on npm), A2AServer adapts a Strands Agent to the Agent2Agent (A2A) protocol. Before Python 1.43.0 (commit 398343faad) and TypeScript 1.5.0 (commit f8ba9376fe), A2AServer accepted a single agent: SAAgent instance and passed that same object into StrandsA2AExecutor.
Because a Strands Agent holds its conversation messages (agent.messages), system prompt state, and session manager on the instance itself (and StrandsA2AExecutor also stored _current_artifact_id and _is_first_chunk on self), every incoming A2A request mutated the exact same conversation history regardless of the request’s context_id:
- Cross-tenant conversation disclosure: Caller B sending a request with a fresh
context_idcaused the model to run with Caller A’s prior turns still inagent.messages. - Cross-session prompt injection: Any caller could plant persistent instructions in the shared
agent.messageslist that steered tool calls and responses for subsequent callers.
The fix introduces agent_factory: Callable[[str], Agent] (agentFactory: (contextId: string) => InvokableAgent in TypeScript) alongside an LRU context map (max_contexts = 1000) where each context_id owns a dedicated _ContextEntry(agent, lock) and per-request _StreamState:
# strands-py/src/strands/multiagent/a2a/server.py (patched in 1.43.0)
from strands import Agent
from strands.multiagent.a2a import A2AServer
# DeprecationWarning if a single `agent=` is passed; use agent_factory for multi-caller isolation:
server = A2AServer(
agent_factory=lambda context_id: Agent(
model=my_model,
tools=[...],
# Wire any session_manager per context_id here
),
max_contexts=1000,
)
If legacy code still passes a single agent= instance, 1.43.0 / 1.5.0 emits a DeprecationWarning, serializes all requests behind a single lock, and swaps take_snapshot() / load_snapshot() state onto and off the shared instance per context_id (provided no shared session_manager is attached). For real concurrency and multi-tenant isolation, switch to agent_factory.
2. LangGraph (CVE-2026-115883): dot-joined namespace collision in PostgresStore and SqliteStore
In LangGraph’s persistence layer, namespaces are passed as tuples of strings such as ("org_123", "user_456"). Both PostgresStore (langgraph-checkpoint-postgres) and SqliteStore (langgraph-checkpoint-sqlite) store that tuple in SQL as a single dot-joined prefix column (".".join(namespace) -> "org_123.user_456").
To prevent ("org_123.user_456",) from colliding with ("org_123", "user_456"), BaseStore._validate_namespace() forbade periods (.) in namespace labels. However, before commit bfcfea554e, BaseStore only called _validate_namespace() inside put() and aput():
store.get(namespace, key),store.search(namespace_prefix),store.delete(namespace, key), andstore.list_namespaces(prefix=..., suffix=...)(plusaget,asearch,adelete,alist_namespaces) never validated the labels.store.batch(ops)andstore.abatch(ops)validated nothing at all, not evenPutOp(namespace, key, value)items passed directly insidebatch().
Consequently, if an application built a namespace tuple from a caller-supplied string tenant_id (for example namespace = (f"tenant_{user_input}",) or (user_input,)), passing "tenant_a.secret_ns" allowed the caller to:
# Against langgraph-checkpoint-postgres < 3.1.3 or langgraph-checkpoint-sqlite < 3.1.2:
store.get(("tenant_a.secret_ns",), "api_keys") # reads ("tenant_a", "secret_ns")
store.delete(("tenant_a.secret_ns",), "api_keys") # deletes ("tenant_a", "secret_ns")
store.batch([PutOp(("tenant_a.secret_ns",), "k", {})]) # overwrites ("tenant_a", "secret_ns")
Commit bfcfea554e adds _validate_namespace_labels() to get, search, delete, list_namespaces, and their async counterparts in langgraph-checkpoint 4.3.0 (before queueing into AsyncBatchedBaseStore._aqueue, so one caller’s invalid label cannot fail concurrent callers sharing the batch), and calls validate_op_namespace(op) inside _group_ops() in both PostgresStore (3.1.3) and SqliteStore (3.1.2):
--- a/libs/checkpoint-postgres/langgraph/store/postgres/base.py
+++ b/libs/checkpoint-postgres/langgraph/store/postgres/base.py
@@ -1386,6 +1387,7 @@ def _group_ops(ops: Iterable[Op]) -> tuple[dict[type, list[tuple[int, Op]]], int
grouped_ops: dict[type, list[tuple[int, Op]]] = defaultdict(list)
tot = 0
for idx, op in enumerate(ops):
+ validate_op_namespace(op)
grouped_ops[type(op)].append((idx, op))
tot += 1
return grouped_ops, tot
3. FastMCP (GHSA-7xgj-8392-5q4r and GHSA-vqqp-x5qm-5qp4): post-transform auth bypass and un-bound M2M OAuth client secrets
Prefect’s fastmcp 4.1.2 and 3.4.10 (commit 85df42fbf2) fix two distinct security flaws:
A. Direct component calls skipped post-transform AuthCheck (GHSA-7xgj-8392-5q4r)
FastMCP allows servers to attach component-level authorization rules (such as auth=restrict_tag("admin", scopes=["admin"])) and also attach transforms (ToolTransform, Transform) that rename components or add tags like tags={"admin"} at the provider, server, or session layer.
When a client called list_tools(), FastMCP ran all transforms first and then filtered the transformed list through _tool_auth_allows(tool), hiding the "admin"-tagged tool from unprivileged callers. However, on the direct lookup path (server.call_tool("tagged_tool"), get_resource(), get_prompt()), FastMCP checked authorization on the untransformed component or returned the transformed component from call_next without re-evaluating _component_auth_allows() after the transform pipeline finished. Furthermore, if a concrete resource (secret://data) was denied by its AuthCheck and a wildcard resource template (secret://{name}) existed on the same server or via a ProxyProvider, falling through on denial allowed the template to serve the protected URI anyway.
The fix in 85df42fbf2 re-runs _component_auth_allows() on the final transformed Tool, Resource, ResourceTemplate, and Prompt at every lookup layer, and ensures a denied concrete resource shadows any matching resource template at that URI.
B. M2M OAuth wrappers omitted issuer binding (GHSA-vqqp-x5qm-5qp4)
In FastMCP 4.0.0a1 through 4.1.1, ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider (fastmcp/client/auth/client_credentials.py) wrapped the upstream mcp Python SDK’s machine-to-machine OAuth providers, which discover the MCP server’s authorization server via RFC 9728 Protected Resource Metadata.
Although the underlying mcp SDK supported an issuer= parameter to verify that the authorization server advertised by the MCP server matched the issuer that actually owns the client’s client_id and client_secret, FastMCP’s wrappers omitted issuer from __init__() and _bind(). If a FastMCP client connected to a user-supplied or compromised MCP server URL, that server could advertise https://attacker[.]example/oauth/token in its metadata and receive the client’s client_id and client_secret in the token-exchange POST.
FastMCP 4.1.2 bumps its dependency floor to mcp>=2.2.0, adds issuer: str | None = None to both providers, isolates cached tokens by (client_id, scopes, issuer), and emits an MCPDeprecationWarning when issuer is omitted:
from fastmcp import Client
from fastmcp.client.auth import ClientCredentialsOAuthProvider
# Upgrading to fastmcp >= 4.1.2 is not enough by itself; pass issuer= explicitly:
auth = ClientCredentialsOAuthProvider(
client_id="my-client-id",
client_secret="my-client-secret",
scopes=["read", "write"],
issuer="https://auth.example.com",
)
4. MLflow (CVE-2026-115886): unauthenticated POST /ajax-api/3.0/mlflow/scorer/invoke SSRF via third_party_scorer_data.kwargs
In MLflow 3.12.0rc0 through 3.16.x, the GenAI evaluation routes POST /ajax-api/3.0/mlflow/scorer/invoke, POST /ajax-api/3.0/mlflow/scorers/register, and POST /ajax-api/3.0/mlflow/genai/evaluate accept a JSON serialized_scorer string. Notably, /ajax-api/3.0/mlflow/scorer/invoke is registered without an authentication decorator on default MLflow tracking servers and runs Scorer.model_validate_json(serialized_scorer) synchronously.
Inside mlflow/genai/scorers/base.py, Scorer.model_validate_json() checked module_path against THIRD_PARTY_SCORER_ALLOWED_MODULES (ragas, deepeval, trulens, phoenix), imported the module, looked up getattr(module, class_name), and called scorer_class(**init_kwargs) with unconstrained data["kwargs"]. When trulens and trulens-providers-litellm were installed in the server environment, selecting module: "mlflow.genai.scorers.trulens" with a non-native model URI such as "openai:/gpt-4o" routed **kwargs directly into trulens.providers.litellm.LiteLLM(model_engine=..., **kwargs), which passed api_base, base_url, api_key, or completion_kwargs into litellm.completion().
Commit bac30b4148 (released in MLflow 3.17.0) adds _validate_serialized_scorer_payload() and _validate_third_party_scorer_data() in mlflow/utils/validation.py, recursively walking every nested third_party_scorer_data dictionary and list of 2-item pairs (used by completion_kwargs) and rejecting any key in GATEWAY_DESTINATION_KEYS (api_base, base_url, api_key, custom_llm_provider, litellm_params, etc.) as well as call_source decorator scorers:
--- a/mlflow/utils/validation.py
+++ b/mlflow/utils/validation.py
@@ -1119,6 +1119,25 @@
+def _find_destination_keys(value: Any) -> set[str]:
+ if isinstance(value, dict):
+ found = GATEWAY_DESTINATION_KEYS & set(value)
+ for nested in value.values():
+ found |= _find_destination_keys(nested)
+ return found
+ if isinstance(value, list):
+ found: set[str] = set()
+ for nested in value:
+ if isinstance(nested, list) and len(nested) == 2 and isinstance(nested[0], str):
+ found |= GATEWAY_DESTINATION_KEYS & {nested[0]}
+ found |= _find_destination_keys(nested)
+ return found
+ return set()
5. MLflow (CVE-2026-116692, CVE-2026-116690, and CVE-2026-116691): third-party scorer artifact RCE, prompt source file read, and CrewAI autolog key leak
Three further MLflow 3.17.0 advisories were published on 27 October:
- Unauthenticated RCE via third-party scorer prefix matching (
CVE-2026-116692/GHSA-26p8-2jq9-3vq9, CVSS 9.8 Critical): Inmlflow/genai/scorers/base.py,Scorer.model_validate()verifiedmodule_pathwith a prefix match:
Becauseif not any( module_path == m or module_path.startswith(m + ".") for m in THIRD_PARTY_SCORER_ALLOWED_MODULES ):mlflow serverallows creating an experiment with a customartifact_location(such asfile:///usr/local/lib/python3.10/site-packages/mlflow/genai/scorers/trulens, which is writable by the defaultuid=0process inghcr.io/mlflow/mlflow:v3.14.0), an unauthenticated attacker could create a run<run_id>, uploadpayload.pyvia the artifact API, and callPOST /ajax-api/3.0/mlflow/scorer/invokewithmodule: "mlflow.genai.scorers.trulens.<run_id>.artifacts.payload".Scorer.model_validate_json()imported the uploaded module and executed its class constructor before checking trace IDs. Commita732443e69replaces the prefix check with an exact membership check (if module_path not in THIRD_PARTY_SCORER_ALLOWED_MODULES:) and validatesclass_nameagainst a closed per-module allowlist before callingimportlib.import_module(). - Arbitrary file read via schemeless prompt model-version
source(CVE-2026-116690/GHSA-m65p-9j7q-6w33, CVSS 7.5 High): When creating a model version carrying themlflow.prompt.is_prompttag,_create_model_version()inmlflow/server/handlers.pyrejectedparsed.scheme == "file"andsource.startswith("/"), but only called_validate_non_local_source_contains_relative_paths(source)whenparsed.schemewas non-empty. Passing a schemeless relative path (source: "../../../etc"or".") stored the traversal path directly as the model version’s artifact root, andGET /api/2.0/mlflow/model-versions/get-artifact?name=pwned-model&version=1&path=passwdserved/etc/passwdorbasic_auth.db. Commit81730a2ce4restricts schemeless prompt sources to_PROMPT_SOURCE_PLACEHOLDERS("dummy-source"and"prompt-template"). - CrewAI autologging API-key disclosure (
CVE-2026-116691/GHSA-j7qq-vr8c-xvw3, CVSS 6.5 Medium): Inmlflow/crewai/autolog.py,_get_agent_attributes()and_construct_full_inputs()expandedAgent.__dict__and convertedagent.llm,agent.function_calling_llm, andagent.agent_executorto strings. Because CrewAI’sLLM.__repr__()includesapi_key='sk-proj-...', enablingmlflow.crewai.autolog()automatically exported plaintext LLM provider API keys into 5 span attributes (attributes.llm,attributes.function_calling_llm,attributes.agent_executor, and twoattributes.mlflow.spanInputsentries). Commit20927930d2reducesLLMinstances to their model name and redacts keys matching_SENSITIVE_KEY_PATTERN.
6. Model Context Protocol Python SDK (mcp 1.30.0 / 2.2.0): CVE-2026-116684, CVE-2026-116686, and CVE-2026-116685
In modelcontextprotocol/python-sdk:
- Opt-in RFC 8707 audience check (
CVE-2026-116684/GHSA-w4fh-qvv9-3v23, commit0c9136841f):BearerAuthBackendinsrc/mcp/server/auth/middleware/bearer_auth.pyaccepted anyAccessTokenreturned bytoken_verifier.verify_token(token)without comparingAccessToken.resourceagainstAuthSettings.resource_server_url. Inmcp1.30.0and2.2.0, you must explicitly opt in by settingvalidate_token_resource=TrueonAuthSettingsand populatingAccessToken.resourceinsideverify_token(). - Tool
outputSchema$refSSRF and file read (CVE-2026-116686/GHSA-rwrf-2pqf-9j8j, commit5ebdfed406): Insrc/mcp/client/session.py,_validate_tool_result()calledjsonschema.validate(result.structuredContent, output_schema)withjsonschema’s default resolver, which fetches unresolved$refURLs viaurllib.request.urlopen()(http://,https://, andfile://) synchronously on the event loop without a timeout.1.30.0and2.2.0pass an explicit emptyregistry=Registry()fromreferencingso only in-document#/$defs/...references resolve. - Same-origin redirect restriction (
CVE-2026-116685/GHSA-5h93-6whr-6q8j, commit2106335745):create_mcp_http_client()no longer enablesfollow_redirects=True;request_within_origin()andstream_within_origin()follow redirects only when the target stays on the same origin (or upgradeshttp -> httpson the same host).
Detection and remediation
1. Audit environments, lockfiles, and vulnerable call patterns
Run the following commands across your Python and TypeScript repositories:
# 1. Check installed Python package versions
pip show strands-agents langgraph-checkpoint langgraph-checkpoint-postgres langgraph-checkpoint-sqlite fastmcp fastmcp-slim mlflow 2>/dev/null || true
# 2. Check npm lockfiles for @strands-agents/sdk < 1.5.0
npm ls @strands-agents/sdk 2>/dev/null || true
# 3. Audit code for single-agent A2AServer(...) construction and FastMCP M2M OAuth without issuer=
rg -n 'A2AServer\s*\(|StrandsA2AExecutor\s*\(|A2AExecutor\s*\(' --glob '*.{py,ts,js}' . 2>/dev/null
rg -n '(ClientCredentialsOAuthProvider|PrivateKeyJWTOAuthProvider)\s*\(' --glob '*.py' . 2>/dev/null
2. Upgrade packages and apply required code changes
- Strands Agents SDK (
CVE-2026-115419):- Upgrade Python
strands-agentsto>= 1.43.0and TypeScript@strands-agents/sdkto>= 1.5.0. - Refactor every multi-caller
A2AServer(agent)/StrandsA2AExecutor(agent)call to passagent_factory(agentFactoryin TypeScript) so eachcontext_idreceives a freshAgentinstance and a context-scopedsession_manager.
- Upgrade Python
- LangGraph (
CVE-2026-115883):- Upgrade
langgraph-checkpoint-postgresto>= 3.1.3,langgraph-checkpoint-sqliteto>= 3.1.2, andlanggraph-checkpointto>= 4.3.0(or upgrade self-hosted LangSmith Agent Server to>= 0.15.5/>= 0.16.2).
- Upgrade
- FastMCP (
GHSA-7xgj-8392-5q4randGHSA-vqqp-x5qm-5qp4):- Upgrade
fastmcp/fastmcp-slimto>= 4.1.2(or>= 3.4.10on the 3.x branch). - Add
issuer="https://<your-expected-auth-server>"to everyClientCredentialsOAuthProviderandPrivateKeyJWTOAuthProviderinstantiation, and rotate anyclient_secretthat was used with untrusted or user-configured MCP server URLs.
- Upgrade
- MLflow (
CVE-2026-115886):- Upgrade
mlflowto>= 3.17.0, and audit tracking-storeregistered_scorersrows and HTTP logs for requests to/ajax-api/3.0/mlflow/scorer/invokecontainingthird_party_scorer_data.
- Upgrade
Timeline
- 2026-10-22 (20:04 UTC): AWS publishes bulletin
2026-146-AWSandCVE-2026-115419(GHSA-4v6f-7vjh-6x4m) alongsidestrands-agents1.43.0and@strands-agents/sdk1.5.0. - 2026-10-23 (21:12 UTC): LangChain publishes
CVE-2026-115883(GHSA-6hmw-9432-67qq) and releaseslanggraph-checkpoint4.3.0,langgraph-checkpoint-postgres3.1.3, andlanggraph-checkpoint-sqlite3.1.2. - 2026-10-23 (21:17 UTC): MLflow publishes
CVE-2026-115886(GHSA-2wqp-q3c6-ccj5), fixed inmlflow3.17.0. - 2026-10-23 (21:31 to 21:32 UTC): Prefect publishes
GHSA-vqqp-x5qm-5qp4andGHSA-7xgj-8392-5q4rforfastmcpandfastmcp-slim, fixed in3.4.10and4.1.2. - 2026-10-27 (21:05 to 21:26 UTC): The GitHub Advisory Database and NVD publish
CVE-2026-116684,CVE-2026-116685, andCVE-2026-116686for the MCP Python SDK (mcp1.30.0and2.2.0) alongsideCVE-2026-116692(CVSS 9.8),CVE-2026-116690, andCVE-2026-116691formlflow3.17.0.
From research to remediation
Check whether this pattern exists in your codebase
Turn this research into a remediation workflow. Scan dependencies and package manifests for similar supply-chain risk, then prioritize fixes with reachability context.
References
- MLflow advisory GHSA-26p8-2jq9-3vq9 and commit a732443e69 (CVE-2026-116692: third-party scorer artifact import RCE)
- Zero Day Initiative advisories ZDI-26-839 (CVE-2026-118623), ZDI-26-843 (CVE-2026-118627), and ZDI-26-844 (CVE-2026-118628) for MLflow
- MLflow advisories GHSA-m65p-9j7q-6w33 (CVE-2026-116690: prompt source file read) and GHSA-j7qq-vr8c-xvw3 (CVE-2026-116691: CrewAI autolog key leak)
- MCP Python SDK advisories GHSA-w4fh-qvv9-3v23 (CVE-2026-116684), GHSA-5h93-6whr-6q8j (CVE-2026-116685), and GHSA-rwrf-2pqf-9j8j (CVE-2026-116686)
- AWS Security Bulletin 2026-146-AWS & GHSA-4v6f-7vjh-6x4m (CVE-2026-115419: Strands Agents SDK A2A session leak)
- strands-agents/harness-sdk commit 398343faad (Python A2A agent_factory isolation)
- LangGraph advisory GHSA-6hmw-9432-67qq (CVE-2026-115883: dot-joined namespace traversal)
- langchain-ai/langgraph commit bfcfea554e (validate namespace labels before store dispatch)
- FastMCP advisory GHSA-7xgj-8392-5q4r (post-transform component authorization skipped on direct calls)
- FastMCP advisory GHSA-vqqp-x5qm-5qp4 (M2M OAuth providers without issuer binding)
- PrefectHQ/fastmcp commit 85df42fbf2 (enforce component authorization and bind OAuth credentials to an issuer)
- MLflow advisory GHSA-2wqp-q3c6-ccj5 (CVE-2026-115886: third-party scorer kwargs SSRF)
- mlflow/mlflow commit bac30b4148 (reject egress-steering kwargs in third-party scorer payloads)
- Strands Agents advisories GHSA-3w55-933p-8pv5 and GHSA-763v-5pjm-fx36 (FileSessionManager /tmp symlink LPE and world-readable session storage, fixed in 1.48.0)
- AWS Security Bulletin 2026-152-AWS & GHSA-4prq-p2x5-w9fh (CVE-2026-115413: AWS API MCP Server OAuth audience bypass, fixed in 1.6.0)
- google/adk-python commit f72f0db58c (CVE-2026-115522: namespace FileArtifactService artifacts by app_name, fixed in 1.39.1 / 2.7.0)
- LangGraph advisories GHSA-4hm6-w6qq-w73v (CVE-2026-118668: langgraph-api JS Store auth bypass), GHSA-3fx2-cqw9-xr3c (CVE-2026-118667), and GHSA-8prr-f582-jqq3 (CVE-2026-115885)
- Mem0 advisory GHSA-5324-3w58-qpp2 and commit 71f2ebefa3 (CVE-2026-116575: _build_session_scope delimiter collision)