🚨 CVE-2026-14696
When Ethernet bridging is enabled (CONFIG_NET_ETHERNET_BRIDGE), eth_bridge_input_process() in subsys/net/l2/ethernet/bridge/bridge_input.c decides how each frame received on a bridge member interface is handled. For frames that must also be delivered to the local stack, the code called eth_bridge_handle_locally() and returned NET_OK. That helper does not consume the packet — it only calls bridge_iface_recv() (via virtual_recv()), which returns NET_CONTINUE without taking ownership of pkt.
The NET_OK verdict then propagates through ethernet_recv() up to processing_data() in subsys/net/ip/net_core.c, where NET_OK is interpreted as "the packet was consumed, do not free it." Because no consumer actually took ownership, the RX net_pkt is never returned to the pool and is leaked. The concretely reproducible leak occurs for frames whose EtherType has no registered L3 handler when CONFIG_NET_ETHERNET_FORWARD_UNRECOGNISED_ETHERTYPE is set (default y when CONFIG_NET_SOCKETS_PACKET is enabled): the fall-through L3 dispatch does not overwrite the NET_OK verdict, so ethernet_recv() returns NET_OK and the buffer is never released.
Any device on a bridged L2 segment can emit broadcast/multicast frames carrying an arbitrary EtherType with no authentication. Each such frame permanently consumes one buffer from the finite RX pool (CONFIG_NET_PKT_RX_COUNT), so a brief broadcast flood exhausts the pool and the device can no longer receive traffic until it is rebooted — a persistent denial of service. There is no confidentiality or integrity impact.
The fix makes eth_bridge_handle_locally() propagate the real net_verdict and return NET_CONTINUE for locally-kept frames, writing the bridge interface back through a new dst_iface out-parameter so the packet follows the normal receive path and is unreferenced exactly once.
🎖@cveNotify
When Ethernet bridging is enabled (CONFIG_NET_ETHERNET_BRIDGE), eth_bridge_input_process() in subsys/net/l2/ethernet/bridge/bridge_input.c decides how each frame received on a bridge member interface is handled. For frames that must also be delivered to the local stack, the code called eth_bridge_handle_locally() and returned NET_OK. That helper does not consume the packet — it only calls bridge_iface_recv() (via virtual_recv()), which returns NET_CONTINUE without taking ownership of pkt.
The NET_OK verdict then propagates through ethernet_recv() up to processing_data() in subsys/net/ip/net_core.c, where NET_OK is interpreted as "the packet was consumed, do not free it." Because no consumer actually took ownership, the RX net_pkt is never returned to the pool and is leaked. The concretely reproducible leak occurs for frames whose EtherType has no registered L3 handler when CONFIG_NET_ETHERNET_FORWARD_UNRECOGNISED_ETHERTYPE is set (default y when CONFIG_NET_SOCKETS_PACKET is enabled): the fall-through L3 dispatch does not overwrite the NET_OK verdict, so ethernet_recv() returns NET_OK and the buffer is never released.
Any device on a bridged L2 segment can emit broadcast/multicast frames carrying an arbitrary EtherType with no authentication. Each such frame permanently consumes one buffer from the finite RX pool (CONFIG_NET_PKT_RX_COUNT), so a brief broadcast flood exhausts the pool and the device can no longer receive traffic until it is rebooted — a persistent denial of service. There is no confidentiality or integrity impact.
The fix makes eth_bridge_handle_locally() propagate the real net_verdict and return NET_CONTINUE for locally-kept frames, writing the bridge interface back through a new dst_iface out-parameter so the packet follows the normal receive path and is unreferenced exactly once.
🎖@cveNotify
GitHub
net: bridge: Fix fatal memory leak in bridge mode. · zephyrproject-rtos/zephyr@4eb007a
When the decision is made to keep the current packet for
further (local) processing it should not return NET_OK as
this will ultimately cause a serious leak of RX packets.
Cleanup logic a bit so w...
further (local) processing it should not return NET_OK as
this will ultimately cause a serious leak of RX packets.
Cleanup logic a bit so w...
🚨 CVE-2026-53507
oasdiff-action is a GitHub Action that detects breaking changes in OpenAPI specs and post a review on every pull request. Before version 0.0.51, the oasdiff actions resolved external $refs in the OpenAPI spec by default (allow-external-refs: true). When an action runs on a pull request whose spec is attacker-controlled — most importantly fork pull requests on public repositories — a $ref in that spec is fetched/read on the runner with no interaction required, enabling SSRF and disclosure of structured files on the runner. This issue has been patched in version 0.0.51.
🎖@cveNotify
oasdiff-action is a GitHub Action that detects breaking changes in OpenAPI specs and post a review on every pull request. Before version 0.0.51, the oasdiff actions resolved external $refs in the OpenAPI spec by default (allow-external-refs: true). When an action runs on a pull request whose spec is attacker-controlled — most importantly fork pull requests on public repositories — a $ref in that spec is fetched/read on the runner with no interaction required, enabling SSRF and disclosure of structured files on the runner. This issue has been patched in version 0.0.51.
🎖@cveNotify
GitHub
harden: default allow-external-refs to false in the actions by reuvenharrison · Pull Request #128 · oasdiff/oasdiff-action
Draft — do not merge until the prerequisites below are met.
What
Makes all five actions default allow-external-refs to false, the safe posture for CI that runs unattended on pull-request (incl. for...
What
Makes all five actions default allow-external-refs to false, the safe posture for CI that runs unattended on pull-request (incl. for...
🚨 CVE-2026-53508
oasdiff is a command-line and Go package that compares and detects breaking changes in OpenAPI specs. From version 1.13.2 through version 1.18.0, oasdiff did not enforce --allow-external-refs=false (library: openapi3.Loader.IsExternalRefsAllowed = false) when loading a spec from a git revision (the rev:path form, e.g. main:openapi.yaml). External $refs were resolved on that load path even when external refs were explicitly disabled, so the mitigation silently did not apply there. This issue has been patched in version 1.18.1.
🎖@cveNotify
oasdiff is a command-line and Go package that compares and detects breaking changes in OpenAPI specs. From version 1.13.2 through version 1.18.0, oasdiff did not enforce --allow-external-refs=false (library: openapi3.Loader.IsExternalRefsAllowed = false) when loading a spec from a git revision (the rev:path form, e.g. main:openapi.yaml). External $refs were resolved on that load path even when external refs were explicitly disabled, so the mitigation silently did not apply there. This issue has been patched in version 1.18.1.
🎖@cveNotify
GitHub
fix: resolve $ref chains via git show when loading from git revision specs by reuvenharrison · Pull Request #832 · oasdiff/oasdiff
Problem
When loading a spec like `origin/main:multi-file/openapi.yaml` that contains `$ref: "./schemas/pet.yaml"`, kin-openapi resolves the relative ref to `origin/main:multi-file...
When loading a spec like `origin/main:multi-file/openapi.yaml` that contains `$ref: "./schemas/pet.yaml"`, kin-openapi resolves the relative ref to `origin/main:multi-file...
🚨 CVE-2026-53552
Goploy is an open-source automation deployment system. In versions 1.17.5 and prior, Project.AddFile, Project.EditFile, Project.RemoveFile, and Project.Edit in cmd/server/api/project/handler.go accept a project or project-file row id from the JSON body and act on it without checking that the project belongs to the caller's namespace. The corresponding model.ProjectFile.GetData and model.Project.GetData queries filter only by row id. A user holding the manager role (or any role that includes the FileSync / EditProject permission) in their own namespace can read, write, or delete files in any project across the install, and can rewrite any project's git remote URL by submitting the foreign id in the body. The git-URL primitive escalates to RCE on the next deploy because Edit runs git remote set-url on the project's working tree. At time of publication, there are no known publicly available patches.
🎖@cveNotify
Goploy is an open-source automation deployment system. In versions 1.17.5 and prior, Project.AddFile, Project.EditFile, Project.RemoveFile, and Project.Edit in cmd/server/api/project/handler.go accept a project or project-file row id from the JSON body and act on it without checking that the project belongs to the caller's namespace. The corresponding model.ProjectFile.GetData and model.Project.GetData queries filter only by row id. A user holding the manager role (or any role that includes the FileSync / EditProject permission) in their own namespace can read, write, or delete files in any project across the install, and can rewrite any project's git remote URL by submitting the foreign id in the body. The git-URL primitive escalates to RCE on the next deploy because Edit runs git remote set-url on the project's working tree. At time of publication, there are no known publicly available patches.
🎖@cveNotify
GitHub
Cross-namespace IDOR and RCE via body-supplied row id in project and project_file handlers
### Summary
`Project.AddFile`, `Project.EditFile`, `Project.RemoveFile`, and `Project.Edit` in `cmd/server/api/project/handler.go` accept a project or project-file row id from the JSON body and ac...
`Project.AddFile`, `Project.EditFile`, `Project.RemoveFile`, and `Project.Edit` in `cmd/server/api/project/handler.go` accept a project or project-file row id from the JSON body and ac...
🚨 CVE-2026-53553
Goploy is an open-source automation deployment system. Prior to version 1.18.0, a severe path traversal vulnerability exists in its backend API endpoints, specifically /deploy/fileDiff (File Compare), when handling file paths provided by the client. This issue has been patched in version 1.18.0.
🎖@cveNotify
Goploy is an open-source automation deployment system. Prior to version 1.18.0, a severe path traversal vulnerability exists in its backend API endpoints, specifically /deploy/fileDiff (File Compare), when handling file paths provided by the client. This issue has been patched in version 1.18.0.
🎖@cveNotify
GitHub
Release v1.18.0 · zhenorzz/goploy
🔒 Security
Prevent path traversal on project file endpoints (GHSA-4g5x-hcwm-82jw)
Scope project & project_file lookups by namespace_id to enforce tenant isolation
Harden path, shell, and repo ...
Prevent path traversal on project file endpoints (GHSA-4g5x-hcwm-82jw)
Scope project & project_file lookups by namespace_id to enforce tenant isolation
Harden path, shell, and repo ...
🚨 CVE-2026-72001
Pangolin before 1.22.0 contains an authentication bypass vulnerability that allows unauthenticated attackers to access any protected resource by supplying an attacker-controlled URL parameter to the share-link authentication endpoint that omits the expected resource identifier from the token verification call. Attackers holding a single valid share link for any resource can authenticate against arbitrary resources across different organizations, bypassing all configured authentication methods including SSO, resource passwords, PIN codes, email allowlists, and header authentication.
🎖@cveNotify
Pangolin before 1.22.0 contains an authentication bypass vulnerability that allows unauthenticated attackers to access any protected resource by supplying an attacker-controlled URL parameter to the share-link authentication endpoint that omits the expected resource identifier from the token verification call. Attackers holding a single valid share link for any resource can authenticate against arbitrary resources across different organizations, bypassing all configured authentication methods including SSO, resource passwords, PIN codes, email allowlists, and header authentication.
🎖@cveNotify
GitHub
Release 1.22.0: Merge pull request #3645 from fosrl/dev · fosrl/pangolin
🚨 CVE-2026-82818
A vulnerability was determined in dibo-software diboot 3.8.0. This affects an unknown part of the file /api/iam/tenant/resource of the component Tenant Resource Assignment Handler. Executing a manipulation of the argument tenantId can lead to improper access controls. The attack may be launched remotely. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way.
🎖@cveNotify
A vulnerability was determined in dibo-software diboot 3.8.0. This affects an unknown part of the file /api/iam/tenant/resource of the component Tenant Resource Assignment Handler. Executing a manipulation of the argument tenantId can lead to improper access controls. The attack may be launched remotely. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way.
🎖@cveNotify
🚨 CVE-2026-82820
A vulnerability was found in FLVMeta up to 1.2.2. Affected is the function amf_string_new of the file src/amf.c of the component AMF String Processing. The manipulation of the argument length results in heap-based buffer overflow. The attack can be launched remotely. The exploit has been made public and could be used. The patch is identified as f412a33b9a84c2d1a9dee145a868feddbf64879e. A patch should be applied to remediate this issue. The project maintainer doubts the security impact: "While I acknowledged the bugs and provided fixes, I have yet to see any way to exploit these alleged vulnerabilities."
🎖@cveNotify
A vulnerability was found in FLVMeta up to 1.2.2. Affected is the function amf_string_new of the file src/amf.c of the component AMF String Processing. The manipulation of the argument length results in heap-based buffer overflow. The attack can be launched remotely. The exploit has been made public and could be used. The patch is identified as f412a33b9a84c2d1a9dee145a868feddbf64879e. A patch should be applied to remediate this issue. The project maintainer doubts the security impact: "While I acknowledged the bugs and provided fixes, I have yet to see any way to exploit these alleged vulnerabilities."
🎖@cveNotify
GitHub
Fix buffer overflow in amf_string_new · noirotm/flvmeta@f412a33
Root issue was that amf_data_clone incorrectly allocated memory for
the string to copy, stopping at the first null character, while AMF
only cares about the declared length of the buffer.
amf_strin...
the string to copy, stopping at the first null character, while AMF
only cares about the declared length of the buffer.
amf_strin...
🚨 CVE-2026-82821
A vulnerability was determined in FLVMeta up to 1.2.2. Affected by this vulnerability is the function amf_object_get of the file src/amf.c of the component AMF Object Parsing. This manipulation causes null pointer dereference. The attack may be initiated remotely. The exploit has been publicly disclosed and may be utilized. Patch name: 52642f7dfb76ec7334016622dde60b1ae963d79b. To fix this issue, it is recommended to deploy a patch. The project maintainer doubts the security impact: "While I acknowledged the bugs and provided fixes, I have yet to see any way to exploit these alleged vulnerabilities."
🎖@cveNotify
A vulnerability was determined in FLVMeta up to 1.2.2. Affected by this vulnerability is the function amf_object_get of the file src/amf.c of the component AMF Object Parsing. This manipulation causes null pointer dereference. The attack may be initiated remotely. The exploit has been publicly disclosed and may be utilized. Patch name: 52642f7dfb76ec7334016622dde60b1ae963d79b. To fix this issue, it is recommended to deploy a patch. The project maintainer doubts the security impact: "While I acknowledged the bugs and provided fixes, I have yet to see any way to exploit these alleged vulnerabilities."
🎖@cveNotify
GitHub
Fix invalid memory access in AMF object iteration · noirotm/flvmeta@52642f7
Invalid AMF payloads can result in the data node being NULL, meaning
we dereference a NULL pointer.
Fixes: #28
we dereference a NULL pointer.
Fixes: #28
🚨 CVE-2026-83497
Unrestricted deserialization of untrusted data in the cursor pagination component in the OpenSearch SQL plugin allows a remote authenticated user with basic read/search permissions to execute arbitrary code on the server by sending a crafted cursor parameter to the plugins/sql endpoint.
🎖@cveNotify
Unrestricted deserialization of untrusted data in the cursor pagination component in the OpenSearch SQL plugin allows a remote authenticated user with basic read/search permissions to execute arbitrary code on the server by sending a crafted cursor parameter to the plugins/sql endpoint.
🎖@cveNotify
🚨 CVE-2026-18064
An incomplete fix for CVE-2026-15352 in the NASA core Flight System
(cFS) Health and Safety (HS) application leaves a separate NULL pointer
dereference reachable in versions through 7.0.1. An attacker who can
trigger the affected command under specific conditions could cause the
HS application to crash, resulting in a denial-of-service condition and
processor reset.
🎖@cveNotify
An incomplete fix for CVE-2026-15352 in the NASA core Flight System
(cFS) Health and Safety (HS) application leaves a separate NULL pointer
dereference reachable in versions through 7.0.1. An attacker who can
trigger the affected command under specific conditions could cause the
HS application to crash, resulting in a denial-of-service condition and
processor reset.
🎖@cveNotify
GitHub
CSAF/csaf_files/OT/white/2026/icsa-26-211-06.json at develop · cisagov/CSAF
CISA CSAF Security Advisories. Contribute to cisagov/CSAF development by creating an account on GitHub.
🚨 CVE-2026-38709
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the net.set_wan interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the net.set_wan interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
Cudy
Cudy-SA-26-7-JEZZY4
A vulnerability allows command execution on the underlying system
🚨 CVE-2026-67822
Tenda W6-S 1.0.0.4(510) contains a stack-based buffer overflow vulnerability in the /goform/wifiSSIDset endpoint. The function formwrlSSIDset uses sprintf to copy user-controlled 'GO' and 'index' parameters into a 64-byte stack buffer without length restriction, leading to stack overflow.
🎖@cveNotify
Tenda W6-S 1.0.0.4(510) contains a stack-based buffer overflow vulnerability in the /goform/wifiSSIDset endpoint. The function formwrlSSIDset uses sprintf to copy user-controlled 'GO' and 'index' parameters into a 64-byte stack buffer without length restriction, leading to stack overflow.
🎖@cveNotify
GitHub
Tenda/Tenda_W6-S_GO_overflow.md at main · Tristerjh/Tenda
Contribute to Tristerjh/Tenda development by creating an account on GitHub.
🚨 CVE-2026-38711
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the system.upgrade_check interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the system.upgrade_check interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
Cudy
Cudy-SA-26-7-7KJW1B
A vulnerability allows command execution on the underlying system
🚨 CVE-2025-69946
SourceCodester Modern Loan Management System 1.0 is vulnerable to SQL Injection in ajaxData.php via the parameters district_id , division_id, region_id, and ward_id.
🎖@cveNotify
SourceCodester Modern Loan Management System 1.0 is vulnerable to SQL Injection in ajaxData.php via the parameters district_id , division_id, region_id, and ward_id.
🎖@cveNotify
GitHub
GitHub - um-dsp/TaintRadar: Multi-Language Code Property Graph-Based Vulnerability Analysis
Multi-Language Code Property Graph-Based Vulnerability Analysis - um-dsp/TaintRadar
🚨 CVE-2025-69948
SourceCodester Modern Loan Management System 1.0 is vulnerable to SQL Injection in /admin/delete_group.php?id=1.
🎖@cveNotify
SourceCodester Modern Loan Management System 1.0 is vulnerable to SQL Injection in /admin/delete_group.php?id=1.
🎖@cveNotify
GitHub
GitHub - um-dsp/TaintRadar: Multi-Language Code Property Graph-Based Vulnerability Analysis
Multi-Language Code Property Graph-Based Vulnerability Analysis - um-dsp/TaintRadar
🚨 CVE-2026-38708
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the system.setclock interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the system.setclock interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
Cudy
Cudy-SA-26-7-AIHGTK
A vulnerability allows command execution on the underlying system
🚨 CVE-2026-38713
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the ipsec_conn interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
TR1200 v2.4.15, TR3000 v2.4.21, WR300 v2.4.25, WR1200 v2.4.23, WR1300 v2.4.22, WR1500 v2.3.10, WR3000 v2.4.19, WR3600 v2.3.16, and WR6500 v2.3.15 were discovered to contain a command injection vulnerability in the ipsec_conn interface. This vulnerability allows attackers to execute arbitrary commands as root via a crafted input.
🎖@cveNotify
Cudy
Cudy-SA-26-7-EHIXBE
A vulnerability allows command execution on the underlying system
🚨 CVE-2026-50986
PrestaShop module, totadministrativemandate <1.8.1 is vulnerable to Cross Site Request Forgery (CSRF). The payment validation controller has no CSRF token. An attacker can confirm an order in an awaiting status by hijacking a link.
🎖@cveNotify
PrestaShop module, totadministrativemandate <1.8.1 is vulnerable to Cross Site Request Forgery (CSRF). The payment validation controller has no CSRF token. An attacker can confirm an order in an awaiting status by hijacking a link.
🎖@cveNotify
Prestashop
Mandat administratif
La solution de paiement la plus adaptée pour les organismes administratifs. Ce module vous propose un moyen de paiement incontournable auprès des administrations et collectivités.
🚨 CVE-2026-51785
An issue in Hugo Leisink Hiawatha v.12.1 and before allows a remote attacker to execute arbitrary code via a crafted request
🎖@cveNotify
An issue in Hugo Leisink Hiawatha v.12.1 and before allows a remote attacker to execute arbitrary code via a crafted request
🎖@cveNotify
Fenrisk
HTTP Request Smuggling in Hiawatha - CVE-2026-51785
Security experts
🚨 CVE-2026-9335
A vulnerability in keras-team/keras versions <= 3.14.0 allows arbitrary local HDF5 file content disclosure due to improper handling of HDF5 ExternalLinks. The `KerasFileEditor` and `keras.saving.load_weights` functions bypass the `safe_get_h5_group` and `safe_get_h5_dataset` helpers, which are designed to reject ExternalLinks and SoftLinks. This results in automatic dereferencing of links to external HDF5 files, enabling attackers to disclose sensitive data from the victim's local filesystem. Specifically, `KerasFileEditor` extracts attributes and datasets from linked files into its internal structures, while `keras.saving.load_weights` loads weights from linked files into the user's model. This issue can be exploited by providing a malicious `.h5`, `.weights.h5`, or `.keras` file containing ExternalLinks.
🎖@cveNotify
A vulnerability in keras-team/keras versions <= 3.14.0 allows arbitrary local HDF5 file content disclosure due to improper handling of HDF5 ExternalLinks. The `KerasFileEditor` and `keras.saving.load_weights` functions bypass the `safe_get_h5_group` and `safe_get_h5_dataset` helpers, which are designed to reject ExternalLinks and SoftLinks. This results in automatic dereferencing of links to external HDF5 files, enabling attackers to disclose sensitive data from the victim's local filesystem. Specifically, `KerasFileEditor` extracts attributes and datasets from linked files into its internal structures, while `keras.saving.load_weights` loads weights from linked files into the user's model. This issue can be exploited by providing a malicious `.h5`, `.weights.h5`, or `.keras` file containing ExternalLinks.
🎖@cveNotify
GitHub
Fix: KerasFileEditor eject HDF5 ExternalLink/SoftLink groups (#22899) · keras-team/keras@23370f1
* KerasFileEditor: reject HDF5 ExternalLink/SoftLink groups
`_extract_weights_from_store` walked the HDF5 tree with raw subscripts
(`data[key]`, `value["vars"]`). When the input ...
`_extract_weights_from_store` walked the HDF5 tree with raw subscripts
(`data[key]`, `value["vars"]`). When the input ...