CVE-2026-77252: MCP Atlassian: filter bypass exposes Jira/Confluence data

MEDIUMPoC AVAILABLE
Published September 22, 2026
CISO Take

MCP Atlassian, the MCP server that lets AI agents query Jira and Confluence, lets callers override the administrator-configured project and space allowlists by supplying their own projects_filter or spaces_filter arguments (CWE-284, CVSS 6.5). Any agent or user able to call the tool can therefore read Jira projects and Confluence spaces that the service credentials can reach but that were meant to be off-limits, which is a direct confidentiality risk for HR, security, legal or M&A content. Exploitation is trivial over the network with low privileges, and a public PoC exists. EPSS is low (0.3%, top 79th percentile), it is not in CISA KEV and SSVC says TRACK, so this is a scheduled-patching item rather than an emergency. Upgrade mcp-atlassian to 0.22.0 or later, and scope the underlying Atlassian API token to least privilege, since the allowlist cannot be relied on as a security boundary in earlier versions.

Sources:NVDEPSSGitHub AdvisoryATLASCISA SSVC

What is the risk?

Medium. The attack is network-reachable, low complexity, needs low privileges and no user interaction, and has high confidentiality impact with no integrity or availability impact. A public PoC exists, but EPSS is low (0.00298), there is no KEV listing, and SSVC is TRACK. The real exposure depends on how much wider the service credentials' access is than the configured allowlist. In agentic deployments the caller may be an LLM steered by prompt injection, which raises the practical likelihood. The package has only 1 downstream dependent, so the supply chain blast radius is small, but it has 38 other CVEs, which suggests a weak security track record.

How does the attack unfold?

Initial Access
The attacker, or an agent steered by prompt injection, gains low-privilege access to call the MCP Atlassian search tools.
AML.T0053
Boundary Bypass
The call supplies projects_filter or spaces_filter, or project/space clauses in the query, which override the administrator allowlist.
AML.T0053
Collection
The server runs the search with the service credentials and returns issues and pages from restricted projects and spaces.
AML.T0085.001
Impact
Confidential data is exposed to the requester or exfiltrated through the agent's output channels.
AML.T0036

What systems are affected?

PackageEcosystemVulnerable RangePatched
MCP Atlassianpip—No patch
2 dependents83% patched~4d to patchFull package profile →

Do you use MCP Atlassian? You're affected.

How severe is it?

CVSS 3.1
6.5 / 10
EPSS
0.3%
chance of exploitation in 30 days
Higher than 23% of all CVEs
Exploitation Status
Exploit Available
Exploitation: MEDIUM
Sophistication
Trivial
Exploitation Confidence
medium
○Public PoC indexed (trickest/cve)
Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVACPRUISCIA
AVNetwork
ACLow
PRLow
UINone
SUnchanged
CHigh
INone
ANone

What should I do?

6 steps
  1. Upgrade mcp-atlassian to 0.22.0 or later (fix in PR #1448, commit b041733).

  2. Until patched, do not rely on JIRA_PROJECTS_FILTER or CONFLUENCE_SPACES_FILTER as a security control; use an Atlassian service account or API token that can only access the intended projects and spaces.

  3. Rotate or re-scope tokens used by the MCP server, following least privilege.

  4. Review MCP server and Atlassian audit logs for search calls that carry projects_filter or spaces_filter values, or JQL/CQL with project or space clauses, outside the allowlist.

  5. Limit which agents and users can invoke the search tools, and require human approval for cross-project queries where feasible.

  6. Inventory MCP servers in developer and agent environments, since these are often deployed without central review.

What does CISA's SSVC say?

DecisionTrack
Exploitationnone
AutomatableNo
Technical Impactpartial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act
Art. 15-Accuracy, robustness and cybersecurity
ISO 42001
A.6.2.6-AI system operation and monitoring
NIST AI RMF
MANAGE 2.3-Mechanisms to respond to and manage unknown risks
OWASP LLM Top 10
LLM02-Sensitive Information DisclosureLLM06-Excessive Agency

Frequently Asked Questions

What is CVE-2026-77252?

MCP Atlassian, the MCP server that lets AI agents query Jira and Confluence, lets callers override the administrator-configured project and space allowlists by supplying their own projects_filter or spaces_filter arguments (CWE-284, CVSS 6.5). Any agent or user able to call the tool can therefore read Jira projects and Confluence spaces that the service credentials can reach but that were meant to be off-limits, which is a direct confidentiality risk for HR, security, legal or M&A content. Exploitation is trivial over the network with low privileges, and a public PoC exists. EPSS is low (0.3%, top 79th percentile), it is not in CISA KEV and SSVC says TRACK, so this is a scheduled-patching item rather than an emergency. Upgrade mcp-atlassian to 0.22.0 or later, and scope the underlying Atlassian API token to least privilege, since the allowlist cannot be relied on as a security boundary in earlier versions.

Is CVE-2026-77252 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-77252, increasing the risk of exploitation.

How to fix CVE-2026-77252?

1. Upgrade mcp-atlassian to 0.22.0 or later (fix in PR #1448, commit b041733). 2. Until patched, do not rely on JIRA_PROJECTS_FILTER or CONFLUENCE_SPACES_FILTER as a security control; use an Atlassian service account or API token that can only access the intended projects and spaces. 3. Rotate or re-scope tokens used by the MCP server, following least privilege. 4. Review MCP server and Atlassian audit logs for search calls that carry projects_filter or spaces_filter values, or JQL/CQL with project or space clauses, outside the allowlist. 5. Limit which agents and users can invoke the search tools, and require human approval for cross-project queries where feasible. 6. Inventory MCP servers in developer and agent environments, since these are often deployed without central review.

What systems are affected by CVE-2026-77252?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, MCP servers, tool-augmented LLM assistants, enterprise knowledge retrieval, RAG pipelines.

What is the CVSS score for CVE-2026-77252?

CVE-2026-77252 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.30%.

What is the AI security impact?

Affected AI Architectures

agent frameworksMCP serverstool-augmented LLM assistantsenterprise knowledge retrievalRAG pipelines

MITRE ATLAS Techniques

AML.T0036Data from Information Repositories
AML.T0051.001Indirect
AML.T0053AI Agent Tool Invocation
AML.T0085.001AI Agent Tools

Compliance Controls Affected

EU AI Act: Art. 15
ISO 42001: A.6.2.6
NIST AI RMF: MANAGE 2.3
OWASP LLM Top 10: LLM02, LLM06

What are the technical details?

Original Advisory

MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, caller-supplied projects_filter and spaces_filter arguments can replace administrator-configured allowlists, and caller-provided project or space clauses can suppress the configured restriction. A caller can search projects or spaces outside the intended boundary when the configured Atlassian credentials can access them. The advisory traces the vulnerable input and processing flow through JIRA_PROJECTS_FILTER, CONFLUENCE_SPACES_FILTER, projects_filter, spaces_filter, SearchMixin.search_issues, and SearchMixin.search, which identify the affected entry points, controls, and code paths. This issue is fixed in version 0.22.0.

Exploitation Scenario

An internal AI assistant uses MCP Atlassian with JIRA_PROJECTS_FILTER limited to the support team's projects, while the shared service token can also read the security-incident and HR projects. A low-privileged user, or an attacker who plants an indirect prompt injection in a Jira ticket or Confluence page the agent reads, makes the agent call the search tool with projects_filter set to the restricted project key, or embeds a project clause in the JQL. The server replaces or ignores the administrator allowlist and returns issues from the restricted projects. The agent then summarizes or relays that content to the requester or to an attacker-controlled channel, leaking confidential data.

Weaknesses (CWE)

CWE-284 — Improper Access Control: The product does not restrict or incorrectly restricts access to a resource from an unauthorized actor.

  • [Architecture and Design, Operation] Very carefully manage the setting, management, and handling of privileges. Explicitly manage trust zones in the software.
  • [Architecture and Design] Compartmentalize the system to have "safe" areas where trust boundaries can be unambiguously drawn. Do not allow sensitive data to go outside of the trust boundary and always be careful when interfacing with a compartment outside of the safe area. Ensure that appropriate compartmentalization is built into the system design, and the compartmentalization allows for and reinforces privilege separation functionality. Architects and designers should rely on the principle of least privilege to decide the appropriate time to use privileges and the time to drop privileges.

Source: MITRE CWE corpus.

CVSS Vector

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N

Timeline

Published
September 22, 2026
Last Modified
September 23, 2026
First Seen
September 22, 2026

Related Vulnerabilities