Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessGain low-privileged team access
-
ExploitSubmit crafted delegation task
-
ExecutionMember agent runs without confirmation
-
ImpactAction executed with excessive privileges
Vulnerability Assessment
| Weakness |
CWE
ASVS 5.0
V8 - Authorization Unnecessary privilege execution
|
| Exploitation Conditions | Requires a deployment using agno's Team Delegation Feature (the delegate_task_to_member / delegate_task_to_members tooling) on affected versions up to 3.0.6/3.1.1, where member-agent execution lacks the requires_confirmation gate. An actor able to influence delegated task content - consistent with the PR:L low-privilege requirement in the CVSS vector - can cause member agents to run actions with unnecessary, unscoped privileges with no human confirmation step. Limiting factors: the team/delegation feature must actually be in use, low-privilege access or task-input influence is needed (not fully anonymous), and the real-world impact is bounded (Low confidentiality/integrity/availability, no subsequent-system scope change). |
| Impact | This is a moderate, defense-in-depth issue rather than a high-priority emergency. The provided CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N) scores only 5.3, with all impacts rated Low (VC:L/VI:L/VA:L) and no scope/subsequent-system change (SC:N/SI:N/SA:N), which is consistent with CWE-250 (execution with unnecessary privileges) - the delegation feature runs member agents without a confirmation gate rather than granting an attacker direct code execution. Exploitation status is 'publicly available exploit code exists' (POC published via a GitHub security advisory and referenced by VulDB); there is NO CISA KEV listing and NO EPSS score in the input, so real-world exploitation frequency cannot be quantified and should not be assumed. PR:L means exploitation requires some low-privilege access/interaction with the team, not fully unauthenticated abuse. Note a data conflict: the input labels the patch 'Available from vendor,' but the description and the referenced pull request (#10124) indicate the fix 'awaits acceptance' - i.e., it is a pending PR, not a confirmed released version; verify the fix is merged/released before relying on it. Overall: a legitimate hardening fix worth applying, but the low impact metrics, PR:L requirement, and absence of KEV/EPSS signals make this a medium-at-most priority, not a critical one. |
| Exploit ScenarioDerived | An attacker with low-privileged access to an agno Team (for example via crafted or prompt-injected task input) triggers the delegation feature so that a member agent is executed without any confirmation or scoping gate. Because member_agent.run proceeds automatically, the delegated action runs with the team's privileges, allowing unintended operations. A public proof-of-concept advisory demonstrating the unscoped/unconfirmed delegation is already available. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 12, 2026
Agno's Team Delegation Feature runs member agents without a confirmation gate, allowing an actor who can influence delegated task content to cause team agents to execute actions with unnecessary, unscoped privileges. The flaw resides in the delegate_task_to_member / delegate_task_to_members tooling in libs/agno/agno/team/_default_tools.py and affects agno versions up to 3.0.6 and 3.1.1. Publicly available exploit code exists (no CISA KEV listing), and the upstream fix - PR #10124, which sets delegate_func.requires_confirmation = True - still awaits acceptance, so no released patched version is independently confirmed.
RemediationAI
No released patched version is confirmed: the fix is contained in upstream pull request https://github.com/agno-agi/agno/pull/10124, which adds delegate_func.requires_confirmation = True to the delegation tool factory, and that PR awaits acceptance/merge. Until a tagged release ships, the most direct compensation is to apply the same control manually by wrapping the delegation factory so the returned delegate_task_to_member/delegate_task_to_members functions carry requires_confirmation = True (mirroring the PR and its test fixture that monkeypatches _default_tools._get_delegate_task_function); be aware this converts delegations into paused, human-in-the-loop runs and will break automated pipelines that consume AG-UI runs without resuming confirmation ToolMessages, so integration code must be updated to handle the PAUSED state and resume flow. Alternatively, disable or avoid the Team Delegation Feature entirely (do not attach delegation tools to teams) at the cost of losing multi-agent delegation - the safest option for unattended deployments. Because exploitation requires low-privilege influence over delegated task content (PR:L), also restrict who can submit or inject task text into team runs and treat delegated content as untrusted input, and monitor team run logs for unexpected delegate_task_to_member executions with unscoped member output. Consult https://vuldb.com/vuln/416597 and https://github.com/Agno-agi/Agno/issues/10054 for updates on the merged fix.
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 an instance of CWE-250 (Execution with Unnecessary Privileges) in the Agno multi-agent framework's team abstraction. When one agent (the team leader) delegates work to a member agent through the default delegation tools, the member's member_agent.run invocation was dispatched without the requires_confirmation safeguard that Agno applies to other confirmation-gated tools. Because delegation inherits the leader's/team's tool privileges and context rather than a scoped, least-privilege capability set, any caller or task input that can steer delegated content can direct a member agent to run actions with the delegation's full authority and no human approval step. The upstream patch is minimal and telling: a single line adding delegate_func.requires_confirmation = True in libs/agno/agno/team/_default_tools.py, plus test updates confirming that the team run now pauses with the requirement tool_execution.tool_name == "delegate_task_to_member" and that private/diagnostic member output no longer leaks into the public payload. Previously, tests asserted the run completed inline with member results exposed, illustrating that delegation executed immediately and transparently. The attack is network-reachable (AV:N) and requires only low-privilege influence over delegation input (PR:L), with no user interaction required (UI:N), matching the assessed CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L profile.
Know when Agno 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:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N/E:P
Remote task input (AV:N/AC:L) but requires low-privilege team interaction (PR:L, no UI); missing confirmation yields limited, bounded C/I/A impact with no scope change.
Primary rating from Vendor (VulDB).
CVSS VectorVendor: VulDB
Lifecycle Timeline
3DescriptionCVE.org
A weakness has been identified in agno-agi agno up to 3.0.6/3.1.1. This issue affects the function member_agent.run in the library libs/agno/agno/team/_default_tools.py of the component Delegation Feature. Executing a manipulation can lead to execution with unnecessary privileges. The attack can be executed remotely. The exploit has been made available to the public and could be used for attacks. The pull request to fix this issue awaits acceptance.
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
Privilege escalation occurs when an attacker leverages flaws in access control mechanisms to gain permissions beyond what they were originally granted.
Model-influenced tool calls in the Agno agent framework (agno-agi) through 3.0.6 and 3.1.1 reach Python exec()/runpy()/s
Missing authentication in Agno (agno-agi) through version 2.7.3 and 3.0.6 causes AgentOS agent endpoints to be reachable
Remote code execution in Agno prior to version 2.3.24 allows attackers to execute arbitrary Python code by manipulating
SQL injection in agno 2.6.5's ClickHouse vector database backend allows authenticated attackers to inject arbitrary SQL
OS command injection in the Agno agent framework's CodingTools.run_shell helper (versions up to and including 3.0.9) let
Code injection in the PythonTools component of Agno (agno-agi/agno) through version 3.0.6 lets an attacker who can influ
EUVD-2026-97319
GHSA-rh58-m2cv-88pv
NVD
CVE.org
EPSS
VulDB
Exploit-DB