Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain low-privileged shell in workspace
-
ExploitRequest 127.0.0.1:2113/debug/manifest
-
ExecutionParse template-provided environment variables
-
ImpactExtract and reuse cloud/registry credentials
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A01:2025 - Broken Access Control
ASVS 5.0
V14 - Data Protection Data exposure prevention
MITRE ATT&CK
2 techniques across 2 tactics: Collection, Discovery The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires a lower-privileged local Unix user or unprivileged process running inside the same Coder workspace as the agent; the agent debug HTTP server must be present and bound to 127.0.0.1:2113 (default in affected 2.10.0-2.34.7 builds), and the workspace template must actually define environment variables containing secrets for there to be anything worth stealing. Limiting factors: the endpoint is loopback-only and therefore NOT remotely reachable, so no network exposure exists; impact is bounded to whatever secrets are present in the template-provided env vars; single-user workspaces with no untrusted co-tenant processes face little practical risk. Fixed in 2.34.8. |
| Impact | This is a genuine but moderate, locally-scoped information-disclosure issue rather than a mass-exploitation priority. The official CVSS 4.0 vector (AV:L/AC:L/AT:N/PR:L/UI:N) correctly models the real constraint: despite the description calling the debug server 'unauthenticated,' exploitation is NOT remote - the server binds to loopback (127.0.0.1:2113), so an attacker must already be a lower-privileged Unix user or process co-located inside the same Coder workspace (hence PR:L, AV:L). The 'unauthenticated' wording refers only to the HTTP endpoint lacking its own auth check, not to network-level unauthenticated access; this nuance should be kept clear to avoid overrating it as a remote flaw. No EPSS score, no CISA KEV listing, and no public POC were provided, so there is no public exploit identified at time of analysis and no evidence of active exploitation - consistent with a bug requiring prior local foothold. The one factor that elevates impact above the base score's 4.8 is the TYPE of data exposed: cloud keys and registry tokens are high-value secrets enabling lateral movement/privilege escalation, which is why my independent assessment rates confidentiality higher (C:H) than the vendor's VC:L. Net: patch it as part of normal maintenance and prioritize on hosts where untrusted or multi-tenant users share workspaces, but it is not a drop-everything remote RCE. |
| Exploit ScenarioDerived | In a shared Coder workspace, a low-privileged local user or a compromised unprivileged process issues an HTTP GET to http://127.0.0.1:2113/debug/manifest and receives the full agent manifest, including template-provided environment variables. The attacker harvests embedded cloud provider keys and container-registry tokens and uses them to access cloud resources or pull/push private images. No public exploit code is identified, but the attack is trivial to script given only shell access inside the workspace. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 17, 2026
Information disclosure in Coder 2.10.0 through 2.34.7 allows a lower-privileged local Unix user or unprivileged process inside the same workspace to read secrets from the workspace agent's debug HTTP server bound to 127.0.0.1:2113, because the /debug/manifest handler returns the complete agent manifest including template-provided environment variable values such as cloud keys and registry tokens. The listener is loopback-only and requires an existing local foothold in the workspace (CVSS v4.0 4.8; independently assessed CVSS 3.1 AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N), so this is a genuine but moderate, locally-scoped issue rather than a mass-exploitation priority, with impact bounded to whatever secrets the template actually injects. No public exploit code or confirmed active exploitation was identified at time of analysis, and the flaw is fixed in 2.34.8.
RemediationAI
Upgrade the Coder control plane and workspace agents to 2.34.8 or later, the version referenced by the vendor advisory (https://github.com/coder/coder/security/advisories/GHSA-73wg-wxvv-9r4j), then restart affected workspaces so rebuilt agents stop serving the unauthenticated /debug/manifest endpoint. Where upgrading immediately is not possible, compensating controls are: disable the agent debug listener if your Coder build exposes a configuration switch for it (this removes the diagnostic capability your support team may rely on for troubleshooting); block 127.0.0.1:2113 with host firewall or network-namespace policy inside the workspace (partial only, since a privileged co-tenant can often bypass loopback restrictions and any policy change may break agent health checks); eliminate the underlying value of the leak by moving long-lived cloud keys and registry tokens out of template-defined environment variables into short-lived, workspace-scoped credentials obtained at runtime (real security benefit but requires template rework and a secrets-management dependency); and restrict workspaces to a single trusted user with no shared or sidecar processes in the same network namespace, which removes the untrusted co-tenant threat model at the cost of multi-user collaboration. Rotate any cloud or registry credentials that were present in template environment variables on affected workspaces, as they should be presumed exposed to any co-tenant process.
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
Coder is a self-hosted remote development platform in which each workspace runs a Go-based agent that manages terminal, port-forwarding, and application sessions on behalf of the control plane. Affected agent builds ship a debug HTTP server that binds to the loopback address 127.0.0.1:2113 and, per this advisory, serves a /debug/manifest endpoint without any authentication or authorization check. That handler serializes the agent's full manifest, which includes environment variables supplied by the workspace template - precisely where platform operators commonly place cloud provider credentials, container registry tokens, and other long-lived secrets so tooling inside the workspace can use them. The root cause is CWE-200 (Exposure of Sensitive Information to an Unauthorized Actor): an internal diagnostic surface is reachable by any process in the same network namespace and discloses data beyond what the requesting principal should see. Because the socket is loopback-bound, the exposure is confined to co-tenants of the workspace host - a second Unix account, a sidecar container sharing the network namespace, or a compromised unprivileged process - and cannot be reached over the network. CPE cpe:2.3:a:coder:coder tracks the affected range, with 2.10.0 as the introduction point and 2.34.8 as the fix.
Know when Coder 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:L/AC:L/AT:N/PR:L/UI:N/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N
Loopback-only endpoint requires a co-located low-privileged local account (AV:L, PR:L); no UI or complexity hurdles; exposes high-value cloud/registry secrets so C:H, with no integrity or availability impact.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
Lifecycle Timeline
1DescriptionCVE.org
Coder from 2.10.0 before 2.34.8 has an unauthenticated agent debug HTTP server on 127.0.0.1:2113 whose /debug/manifest handler returns the full agent manifest, including template-provided environment variable values. A lower-privileged Unix user or process in the same workspace can query this endpoint to read credentials such as cloud keys or registry tokens.
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
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
Coder allows organizations to provision remote development environments via Terraform. Rated high severity (CVSS 8.1), t
Coder allows organizations to provision remote development environments via Terraform. Prior to 2.26.5, 2.27.7, and 2.28
code-server is vulnerable to Inefficient Regular Expression Complexity. Rated high severity (CVSS 7.5), this vulnerabili
Versions of the package code-server before 4.10.1 are vulnerable to Missing Origin Validation in WebSockets handshakes.
Credential theft in Coder (before 2.29.20, 2.33.0 through 2.33.11, 2.34.0 through 2.34.7, and 2.35.0 through 2.35.3) let
Authenticated Coder users who control a modified workspace agent can redirect the control plane's workspace-agent API re
Coder deployments that run workspace agents on Google Compute Engine with GCP instance-identity authentication accept re
Coder allows oragnizations to provision remote development environments via Terraform. Rated high severity (CVSS 8.2), t
Coder self-hosted deployments expose an unauthenticated account-enumeration oracle that lets remote attackers distinguis
IP spoofing in the Coder platform (versions before 2.34.7) allows unauthenticated remote attackers to bypass per-IP rate
Cross-organization file disclosure in Coder's provisioner daemon (versions 2.30.0 through 2.34.7) lets an authenticated
Information disclosure in Coder 2.26.0 through 2.34.7 lets authenticated users who have been granted workspace 'use' acc
EUVD-2026-103293
GHSA-3gj5-9m6g-289w
NVD
CVE.org
EPSS
Vulnrichment
VulDB
Exploit-DB