Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessReach exposed /chat endpoint
-
DeliverySend unauthenticated chat request
-
ExploitInstruct agent to call exec_shell
-
ExecutionRun arbitrary command via subprocess.run
-
ImpactExfiltrate environment secrets and keys
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A07:2025 - Authentication Failures
ASVS 5.0
V6 - Authentication Authentication required for critical functions
MITRE ATT&CK
1 technique across 1 tactic: Initial Access The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires a piflow 2.0.0-2.0.1 deployment running the DeepAgents chat agent with its POST /chat and /chat/stream endpoints network-reachable by the attacker; these endpoints lack authentication by design, and the exec_shell tool executes commands with no allow-list and no human-approval gate, so exploitation works against the default agent configuration once the feature is exposed. No credentials, user interaction, or special privileges are needed (PR:N/UI:N). The practical limiting factors are: the chat agent feature must actually be enabled and exposed to the attacker's network position, the instance must be within the vulnerable 2.0.0-2.0.1 range, and the severity of secret exposure depends on which LLM API keys and database credentials are present in the process environment. |
| Impact | This is a genuine high-priority issue, not a CVSS-inflated paper risk. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:N/UI:N) describes network-reachable, low-complexity, fully unauthenticated exploitation requiring no user interaction, and the impact metrics VC:H/VI:H/VA:H reflect total compromise of the host's confidentiality, integrity, and availability - consistent with CWE-306 (Missing Authentication for Critical Function) and the described arbitrary command execution via subprocess.run. Publicly available exploit code exists (per the Lasso Security reference tagged as an exploit) and the flaw was reported by VulnCheck, which raises practical exploitability above a theoretical finding. No EPSS score and no CISA KEV listing were provided in the input, so there is no evidence of confirmed active exploitation or a quantified exploitation-probability signal - the presence of a POC is a distinct and weaker signal than in-the-wild use. The signals are internally consistent (network vector, no auth, RCE) with no contradictions between the CVSS vector and the description; the main unknown is real-world exposure, since this depends on how many piflow instances run the DeepAgents chat agent on a reachable network. Net: treat as critical-to-patch wherever the affected feature is exposed, but the lack of EPSS/KEV data means widespread automated exploitation is unconfirmed rather than ruled out. |
| Exploit ScenarioDerived | An attacker who can reach a piflow 2.0.0-2.0.1 instance sends an unauthenticated POST request to /chat or /chat/stream, instructing the DeepAgents chat agent to invoke its exec_shell tool. Because the tool runs commands through subprocess.run with no allow-list or human approval, the attacker executes arbitrary OS commands and reads environment secrets such as LLM API keys and database credentials. Public exploit code is referenced, lowering the skill and effort required to weaponize this. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 20, 2026
Unauthenticated remote code execution in PiFlow 2.0.0-2.0.1 allows network attackers to drive the DeepAgents chat agent through POST /chat or POST /chat/stream and invoke the exec_shell tool, which runs arbitrary commands via subprocess.run without human approval or an allow-list. Publicly available exploit code exists, and successful exploitation can expose LLM API keys and database credentials present in the process environment. This is a genuine high-priority risk for exposed deployments where the chat agent feature is enabled and reachable, with no CISA KEV entry indicated in the supplied data.
RemediationAI
No vendor-released patch version is identified in the supplied intelligence; apply the vendor update when published and confirm the fixed release is later than 2.0.1. Until then, block POST /chat and POST /chat/stream at the reverse proxy or WAF (trade-off: disables the DeepAgents chat feature), disable the DeepAgents chat agent if it is not required, and remove or tightly sandbox the exec_shell tool with a strict allow-list and human approval (trade-off: may break legitimate agent workflows and shell-dependent automation). Run PiFlow with a least-privilege container or user, keep LLM API keys and database credentials out of the process environment or mount them only when needed, and rotate any secrets that may have been exposed. Consult https://www.vulncheck.com/advisories/piflow-2.0.0-through-2.0.1-unauthenticated-rce-via-chat-agent-exec-shell and the GitHub source references for details.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
The issue is classified as CWE-306, Missing Authentication for Critical Function, because the DeepAgents chat agent exposes POST /chat and POST /chat/stream without authentication by design. The affected code paths are referenced in the cas-bigdatalab/piflow repository at agents/tools.py line 114 and server.py line 615, where the exec_shell tool executes commands through subprocess.run with no human-approval gate and no command allow-list. The CPE cpe:2.3:a:cas-bigdatalab:piflow:*:*:*:*:*:*:*:* identifies PiFlow as the affected package. The supplied CVSS 4.0 score is 9.3 with a vector showing PR:N/UI:N, and the independent assessment maps this to CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H, consistent with unauthenticated network-reachable code execution that can compromise confidentiality, integrity, and availability. Exploitation requires the chat agent to be enabled and the endpoints to be reachable by the attacker; the severity of secret exposure depends on which LLM API keys and database credentials are present in the process environment.
Know when Piflow gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N
Unauthenticated network endpoint (AV:N/PR:N/UI:N) yielding arbitrary command execution and secret disclosure gives AC:L and C/I/A:H; impact stays on the vulnerable host so S:U.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
DescriptionCVE.org
piflow 2.0.0 through 2.0.1 contains a remote code execution vulnerability in its DeepAgents chat agent, whose POST /chat and POST /chat/stream endpoints lack authentication. Attackers can instruct the agent to call the exec_shell tool, which runs arbitrary commands via subprocess.run without human approval or an allow-list, exposing environment secrets such as LLM API keys and database credentials.
Technical POC Analysis
CWE-306 - Missing Authentication, combined with an unsandboxed tool.
Full technical POC analysis - root cause, step-by-step exploitation and patch breakdown - is part of the Pro and Business plans.
Executive ViewAI
A critical flaw in PiFlow's DeepAgents chat feature lets anyone who can reach the chat endpoints run arbitrary commands on the underlying server, and working exploit code is already circulating. Because the compromised process also holds LLM API keys and database credentials, a single successful attack can cascade into data theft and broader compromise of connected systems. Leadership should care because internet-facing PiFlow deployments are directly exposed, remediation options are limited without a vendor fix, and exploitation requires no special tooling.
No vendor-released patch identified at time of analysis. Interim mitigations: block or restrict network access to POST /chat and POST /chat/stream, disable the DeepAgents chat feature or the exec_shell tool where not essential, run PiFlow with least-privilege service accounts and scoped short-lived credentials, remove secrets from the process environment in favor of a secrets manager, enforce egress filtering to limit post-exploitation activity, and monitor for anomalous subprocess execution and unusual chat endpoint traffic.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Authentication bypass attacks exploit flaws in the verification mechanisms that control access to systems and applications.
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
EUVD-2026-104682
GHSA-9925-27m4-6rvp
NVD
CVE.org
EPSS
Vulnrichment
Exploit-DB